3cf8564...master · text

git.iko.soy/text

78 files changed+3011−1318

since splitsnapshots

unifiedsplit

.gitignorenew file+3

@@ -0,0 +1,3 @@

1+dist

2+cabal-dev

3+cabal.sandbox.config

.hgtags+22

@@ -42,3 +42,25 @@ b75d3041d275e8e76b26605b0d426a572ddf1249 0.11.1.3

42424297307ebc11ad677cfba6b40319e7e5e2c0cfee 0.11.2.3

43437fa79662b66aade97fe49394977213fe6432942e 0.11.3.0

4444d99cd091cdf71ce807a4255f6cc509c3154f51ea 0.11.3.1

45+31f57612fd425d1bb56352ce6b79e41ce6699a45 1.0.0.0

46+7cba97c86467c1881dfced68613891f3c65561b0 1.0.0.1

47+62674a9bbc83145d1d7be28a05d456ab6bb5ff89 1.1.0.0

48+68a32d80c0829da164d1e6b3acedcc3efeda0957 1.1.0.1

49+6d863daafbf0e1289ec0a0eb9b4be2a0a9cb8d3d 1.1.1.0

50+fb7299f22449750c653ac64dee4bd63c13ea6e4c 1.1.1.1

51+eb170cb533b63534d166b5fd5344b36676adf140 1.1.1.2

52+e749e05f715d02445803f653cd3d97d5395d7c6f 1.1.1.3

53+b0dff8ca1142595bab489a768ca62e6a6805aac4 1.1.1.4

54+f8a60279237c1c28e094868740d9188dd94fcc9e 1.2.0.0

55+6231451979b2d33e28038b75f2dab19c9731f5c3 1.2.0.1

56+99a3503b4ad001fc1894e90286e96d7dea1e9195 1.2.0.2

57+47f138edf15576b9adaeed64784e30811099f202 1.2.0.3

58+0b064b0c69106bbcd13cb0099e72ae5fa2347bf0 1.2.0.4

59+8bbb3e222a0037035d893f731b0a23be449f7346 1.2.0.5

60+18c7f8e19872762b7e28ee1871e72f70ea54e3cd 1.2.0.6

61+42f751b87b4a187db93f595a8166d51746ff57d4 1.2.1.0

62+78227a84740b70dcd4cf7a65f3bcf603d87e48e2 1.2.1.1

63+40f268e6735749ef36f487e1d928eaa957bb431d 1.2.1.2

64+631fd055e27fa10983e17e9ad36d1b285b85c661 1.2.1.3

65+8221a68ed5ec37028357df9c6c25f52320e54ad9 1.2.2.0

66+ce1c5ac65eb9f5580114b4e380092b05a485e95d 1.2.2.1

Data/Text.hs+277−166

@@ -3,6 +3,9 @@

33#if __GLASGOW_HASKELL__ >= 702

44{-# LANGUAGE Trustworthy #-}

55#endif

6+#if __GLASGOW_HASKELL__ >= 708

7+{-# LANGUAGE TypeFamilies #-}

8+#endif

710-- |

811-- Module : Data.Text

@@ -11,8 +14,7 @@

1215--

1316-- License : BSD-style

14−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

15−-- duncan@haskell.org

17+-- Maintainer : bos@serpentine.com

1618-- Stability : experimental

1719-- Portability : GHC

1820--

@@ -30,8 +32,8 @@

3032--

3133-- To use an extended and very rich family of functions for working

3234-- with Unicode text (including normalization, regular expressions,

33−-- non-standard encodings, text breaking, and locales), see the

34−-- @text-icu@ package: <http://hackage.haskell.org/package/text-icu>

35+-- non-standard encodings, text breaking, and locales), see

36+-- <http://hackage.haskell.org/package/text-icu the text-icu package >.

3537

3638module Data.Text

3739 (

@@ -73,7 +75,6 @@ module Data.Text

7375 , transpose

7476 , reverse

7577 , replace

76− , replace2

7778

7879 -- ** Case conversion

7980 -- $case

@@ -124,8 +125,11 @@ module Data.Text

124125

125126 -- ** Breaking strings

126127 , take

128+ , takeEnd

127129 , drop

130+ , dropEnd

128131 , takeWhile

132+ , takeWhileEnd

129133 , dropWhile

130134 , dropWhileEnd

131135 , dropAround

@@ -188,82 +192,94 @@ module Data.Text

188192

189193 -- * Low level operations

190194 , copy

195+ , unpackCString#

191196 ) where

192197

193198import Prelude (Char, Bool(..), Int, Maybe(..), String,

194199 Eq(..), Ord(..), Ordering(..), (++),

195− Read(..), Show(..),

196− (&&), (||), (+), (-), (.), ($), ($!), (>>), (*),

197− maxBound, not, return, otherwise, quot)

200+ Read(..),

201+ (&&), (||), (+), (-), (.), ($), ($!), (>>),

202+ not, return, otherwise, quot)

198203#if defined(HAVE_DEEPSEQ)

199−import Control.DeepSeq (NFData)

204+import Control.DeepSeq (NFData(rnf))

200205#endif

201206#if defined(ASSERTS)

202207import Control.Exception (assert)

203208#endif

204209import Data.Char (isSpace)

205−import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf))

206−import Data.Data (mkNoRepType)

210+import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,

211+ Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))

207212import Control.Monad (foldM)

213+import Control.Monad.ST (ST)

208214import qualified Data.Text.Array as A

209215import qualified Data.List as L

216+import Data.Binary (Binary(get, put))

210217import Data.Monoid (Monoid(..))

218+#if MIN_VERSION_base(4,9,0)

219+import Data.Semigroup (Semigroup(..))

220+#endif

211221import Data.String (IsString(..))

212−import qualified Data.Text.Fusion as S

213−import qualified Data.Text.Fusion.Common as S

214−import Data.Text.Fusion (stream, reverseStream, unstream)

215−import Data.Text.Private (span_)

216−import Data.Text.Internal (Text(..), empty, firstf, safe, text, textP)

222+import qualified Data.Text.Internal.Fusion as S

223+import qualified Data.Text.Internal.Fusion.Common as S

224+import Data.Text.Encoding (decodeUtf8', encodeUtf8)

225+import Data.Text.Internal.Fusion (stream, reverseStream, unstream)

226+import Data.Text.Internal.Private (span_)

227+import Data.Text.Internal (Text(..), empty, firstf, mul, safe, text)

228+import Data.Text.Show (singleton, unpack, unpackCString#)

217229import qualified Prelude as P

218230import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,

219− unsafeHead, unsafeTail)

220−import Data.Text.UnsafeChar (unsafeChr)

221−import qualified Data.Text.Util as U

222−import qualified Data.Text.Encoding.Utf16 as U16

223−import Data.Text.Search (indices)

231+ reverseIter_, unsafeHead, unsafeTail)

232+import Data.Text.Internal.Unsafe.Char (unsafeChr)

233+import qualified Data.Text.Internal.Functions as F

234+import qualified Data.Text.Internal.Encoding.Utf16 as U16

235+import Data.Text.Internal.Search (indices)

224236#if defined(__HADDOCK__)

225237import Data.ByteString (ByteString)

226238import qualified Data.Text.Lazy as L

227239import Data.Int (Int64)

228240#endif

229−#if __GLASGOW_HASKELL__ >= 702

230−import qualified GHC.CString as GHC

231−#else

232−import qualified GHC.Base as GHC

241+import GHC.Base (eqInt, neInt, gtInt, geInt, ltInt, leInt)

242+#if __GLASGOW_HASKELL__ >= 708

243+import qualified GHC.Exts as Exts

244+#endif

245+#if MIN_VERSION_base(4,7,0)

246+import Text.Printf (PrintfArg, formatArg, formatString)

233247#endif

234−import GHC.Prim (Addr#)

235248

236249-- $strict

237250--

238251-- This package provides both strict and lazy 'Text' types. The

239−-- strict type is provided by the 'Data.Text' package, while the lazy

240−-- type is provided by the 'Data.Text.Lazy' package. Internally, the

252+-- strict type is provided by the "Data.Text" module, while the lazy

253+-- type is provided by the "Data.Text.Lazy" module. Internally, the

241254-- lazy @Text@ type consists of a list of strict chunks.

242255--

243256-- The strict 'Text' type requires that an entire string fit into

244−-- memory at once. The lazy @Text@ type is capable of streaming

245−-- strings that are larger than memory using a small memory footprint.

246−-- In many cases, the overhead of chunked streaming makes the lazy

247−-- @Text@ type slower than its strict counterpart, but this is not

248−-- always the case. Sometimes, the time complexity of a function in

249−-- one module may be different from the other, due to their differing

250−-- internal structures.

257+-- memory at once. The lazy 'Data.Text.Lazy.Text' type is capable of

258+-- streaming strings that are larger than memory using a small memory

259+-- footprint. In many cases, the overhead of chunked streaming makes

260+-- the lazy 'Data.Text.Lazy.Text' type slower than its strict

261+-- counterpart, but this is not always the case. Sometimes, the time

262+-- complexity of a function in one module may be different from the

263+-- other, due to their differing internal structures.

251264--

252265-- Each module provides an almost identical API, with the main

253266-- difference being that the strict module uses 'Int' values for

254−-- lengths and counts, while the lazy module uses 'Int64' lengths.

267+-- lengths and counts, while the lazy module uses 'Data.Int.Int64'

268+-- lengths.

255269

256270-- $replacement

257271--

258272-- A 'Text' value is a sequence of Unicode scalar values, as defined

259−-- in §3.9, definition D76 of the Unicode 5.2 standard:

260−-- <http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=35>. As

261−-- such, a 'Text' cannot contain values in the range U+D800 to U+DFFF

262−-- inclusive. Haskell implementations admit all Unicode code points

263−-- (§3.4, definition D10) as 'Char' values, including code points

264−-- from this invalid range. This means that there are some 'Char'

265−-- values that are not valid Unicode scalar values, and the functions

266−-- in this module must handle those cases.

273+-- in

274+-- <http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=35 §3.9, definition D76 of the Unicode 5.2 standard >.

275+-- As such, a 'Text' cannot contain values in the range U+D800 to

276+-- U+DFFF inclusive. Haskell implementations admit all Unicode code

277+-- points

278+-- (<http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=13 §3.4, definition D10 >)

279+-- as 'Char' values, including code points from this invalid range.

280+-- This means that there are some 'Char' values that are not valid

281+-- Unicode scalar values, and the functions in this module must handle

282+-- those cases.

267283--

268284-- Within this module, many functions construct a 'Text' from one or

269285-- more 'Char' values. Those functions will substitute 'Char' values

@@ -277,8 +293,8 @@ import GHC.Prim (Addr#)

277293-- range U+D800 through U+DFFF are used by UTF-16 to denote surrogate

278294-- code points, and so cannot be represented. The functions replace

279295-- invalid scalar values, instead of dropping them, as a security

280−-- measure. For details, see Unicode Technical Report 36, §3.5:

281−-- <http://unicode.org/reports/tr36/#Deletion_of_Noncharacters>)

296+-- measure. For details, see

297+-- <http://unicode.org/reports/tr36/#Deletion_of_Noncharacters Unicode Technical Report 36, §3.5 >.)

282298

283299-- $fusion

284300--

@@ -296,11 +312,11 @@ import GHC.Prim (Addr#)

296312-- > countChars = T.length . T.toUpper . E.decodeUtf8

297313--

298314-- From the type signatures involved, this looks like it should

299−-- allocate one 'ByteString' value, and two 'Text' values. However,

300−-- when a module is compiled with optimisation enabled under GHC, the

301−-- two intermediate 'Text' values will be optimised away, and the

302−-- function will be compiled down to a single loop over the source

303−-- 'ByteString'.

315+-- allocate one 'Data.ByteString.ByteString' value, and two 'Text'

316+-- values. However, when a module is compiled with optimisation

317+-- enabled under GHC, the two intermediate 'Text' values will be

318+-- optimised away, and the function will be compiled down to a single

319+-- loop over the source 'Data.ByteString.ByteString'.

304320--

305321-- Functions that can be fused by the compiler are documented with the

306322-- phrase \"Subject to fusion\".

@@ -314,41 +330,82 @@ instance Eq Text where

314330instance Ord Text where

315331 compare = compareText

316332

317−instance Show Text where

318− showsPrec p ps r = showsPrec p (unpack ps) r

319−

320333instance Read Text where

321334 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]

322335

336+#if MIN_VERSION_base(4,9,0)

337+-- Semigroup orphan instances for older GHCs are provided by

338+-- 'semigroups` package

339+

340+instance Semigroup Text where

341+ (<>) = append

342+#endif

343+

323344instance Monoid Text where

324345 mempty = empty

346+#if MIN_VERSION_base(4,9,0)

347+ mappend = (<>) -- future-proof definition

348+#else

325349 mappend = append

350+#endif

326351 mconcat = concat

327352

328353instance IsString Text where

329354 fromString = pack

330355

356+#if __GLASGOW_HASKELL__ >= 708

357+instance Exts.IsList Text where

358+ type Item Text = Char

359+ fromList = pack

360+ toList = unpack

361+#endif

362+

331363#if defined(HAVE_DEEPSEQ)

332−instance NFData Text

364+instance NFData Text where rnf !_ = ()

333365#endif

334366

335−-- This instance preserves data abstraction at the cost of inefficiency.

367+instance Binary Text where

368+ put t = put (encodeUtf8 t)

369+ get = do

370+ bs <- get

371+ case decodeUtf8' bs of

372+ P.Left exn -> P.fail (P.show exn)

373+ P.Right a -> P.return a

374+

375+-- | This instance preserves data abstraction at the cost of inefficiency.

336376-- We omit reflection services for the sake of data abstraction.

337377--

338−-- This instance was created by copying the behavior of Data.Set and

339−-- Data.Map. If you feel a mistake has been made, please feel free to

340−-- submit improvements.

378+-- This instance was created by copying the updated behavior of

379+-- @"Data.Set".@'Data.Set.Set' and @"Data.Map".@'Data.Map.Map'. If you

380+-- feel a mistake has been made, please feel free to submit

381+-- improvements.

341382--

342−-- Original discussion is archived here:

343−

344−-- "could we get a Data instance for Data.Text.Text?"

345−-- http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93

383+-- The original discussion is archived here:

384+-- <http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93 could we get a Data instance for Data.Text.Text? >

385+--

386+-- The followup discussion that changed the behavior of 'Data.Set.Set'

387+-- and 'Data.Map.Map' is archived here:

388+-- <http://markmail.org/message/trovdc6zkphyi3cr#query:+page:1+mid:a46der3iacwjcf6n+state:results Proposal: Allow gunfold for Data.Map, ... >

346389

347390instance Data Text where

348391 gfoldl f z txt = z pack `f` (unpack txt)

349− toConstr _ = P.error "Data.Text.Text.toConstr"

350− gunfold _ _ = P.error "Data.Text.Text.gunfold"

351− dataTypeOf _ = mkNoRepType "Data.Text.Text"

392+ toConstr _ = packConstr

393+ gunfold k z c = case constrIndex c of

394+ 1 -> k (z pack)

395+ _ -> P.error "gunfold"

396+ dataTypeOf _ = textDataType

397+

398+#if MIN_VERSION_base(4,7,0)

399+-- | Only defined for @base-4.7.0.0@ and later

400+instance PrintfArg Text where

401+ formatArg txt = formatString $ unpack txt

402+#endif

403+

404+packConstr :: Constr

405+packConstr = mkConstr textDataType "pack" [] Prefix

406+

407+textDataType :: DataType

408+textDataType = mkDataType "Data.Text.Text" [packConstr]

352409

353410-- | /O(n)/ Compare two 'Text' values lexicographically.

354411compareText :: Text -> Text -> Ordering

@@ -373,30 +430,6 @@ pack :: String -> Text

373430pack = unstream . S.map safe . S.streamList

374431{-# INLINE [1] pack #-}

375432

376−-- | /O(n)/ Convert a Text into a String. Subject to fusion.

377−unpack :: Text -> String

378−unpack = S.unstreamList . stream

379−{-# INLINE [1] unpack #-}

380−

381−-- | /O(n)/ Convert a literal string into a Text. Subject to fusion.

382−unpackCString# :: Addr# -> Text

383−unpackCString# addr# = unstream (S.streamCString# addr#)

384−{-# NOINLINE unpackCString# #-}

385−

386−{-# RULES "TEXT literal" forall a.

387− unstream (S.map safe (S.streamList (GHC.unpackCString# a)))

388− = unpackCString# a #-}

389−

390−{-# RULES "TEXT literal UTF8" forall a.

391− unstream (S.map safe (S.streamList (GHC.unpackCStringUtf8# a)))

392− = unpackCString# a #-}

393−

394−-- | /O(1)/ Convert a character into a Text. Subject to fusion.

395−-- Performs replacement on invalid scalar values.

396−singleton :: Char -> Text

397−singleton = unstream . S.singleton . safe

398−{-# INLINE [1] singleton #-}

399−

400433-- -----------------------------------------------------------------------------

401434-- * Basic functions

402435

@@ -427,12 +460,13 @@ append a@(Text arr1 off1 len1) b@(Text arr2 off2 len2)

427460 | otherwise = overflowError "append"

428461 where

429462 len = len1+len2

463+ x :: ST s (A.MArray s)

430464 x = do

431465 arr <- A.new len

432466 A.copyI arr 0 arr1 off1 len1

433467 A.copyI arr len1 arr2 off2 len

434468 return arr

435−{-# INLINE append #-}

469+{-# NOINLINE append #-}

436470

437471{-# RULES

438472"TEXT append -> fused" [~1] forall t1 t2.

@@ -452,8 +486,8 @@ head t = S.head (stream t)

452486uncons :: Text -> Maybe (Char, Text)

453487uncons t@(Text arr off len)

454488 | len <= 0 = Nothing

455− | otherwise = Just (c, textP arr (off+d) (len-d))

456− where Iter c d = iter t 0

489+ | otherwise = Just $ let !(Iter c d) = iter t 0

490+ in (c, text arr (off+d) (len-d))

457491{-# INLINE [1] uncons #-}

458492

459493-- | Lifted from Control.Arrow and specialized.

@@ -483,7 +517,7 @@ last (Text arr off len)

483517tail :: Text -> Text

484518tail t@(Text arr off len)

485519 | len <= 0 = emptyError "tail"

486− | otherwise = textP arr (off+d) (len-d)

520+ | otherwise = text arr (off+d) (len-d)

487521 where d = iter_ t 0

488522{-# INLINE [1] tail #-}

489523

@@ -498,8 +532,8 @@ tail t@(Text arr off len)

498532-- be non-empty. Subject to fusion.

499533init :: Text -> Text

500534init (Text arr off len) | len <= 0 = emptyError "init"

501− | n >= 0xDC00 && n <= 0xDFFF = textP arr off (len-2)

502− | otherwise = textP arr off (len-1)

535+ | n >= 0xDC00 && n <= 0xDFFF = text arr off (len-2)

536+ | otherwise = text arr off (len-1)

503537 where

504538 n = A.unsafeIndex arr (off+len-1)

505539{-# INLINE [1] init #-}

@@ -538,7 +572,9 @@ isSingleton = S.isSingleton . stream

538572-- Subject to fusion.

539573length :: Text -> Int

540574length t = S.length (stream t)

541−{-# INLINE length #-}

575+{-# INLINE [0] length #-}

576+-- length needs to be phased after the compareN/length rules otherwise

577+-- it may inline before the rules have an opportunity to fire.

542578

543579-- | /O(n)/ Compare the count of characters in a 'Text' to a number.

544580-- Subject to fusion.

@@ -557,32 +593,32 @@ compareLength t n = S.compareLengthI (stream t) n

557593

558594{-# RULES

559595"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.

560− (==) (length t) n = compareLength t n == EQ

596+ eqInt (length t) n = compareLength t n == EQ

561597 #-}

562598

563599{-# RULES

564600"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.

565− (/=) (length t) n = compareLength t n /= EQ

601+ neInt (length t) n = compareLength t n /= EQ

566602 #-}

567603

568604{-# RULES

569605"TEXT <N/length -> compareLength/==LT" [~1] forall t n.

570− (<) (length t) n = compareLength t n == LT

606+ ltInt (length t) n = compareLength t n == LT

571607 #-}

572608

573609{-# RULES

574610"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.

575− (<=) (length t) n = compareLength t n /= GT

611+ leInt (length t) n = compareLength t n /= GT

576612 #-}

577613

578614{-# RULES

579615"TEXT >N/length -> compareLength/==GT" [~1] forall t n.

580− (>) (length t) n = compareLength t n == GT

616+ gtInt (length t) n = compareLength t n == GT

581617 #-}

582618

583619{-# RULES

584620"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.

585− (>=) (length t) n = compareLength t n /= LT

621+ geInt (length t) n = compareLength t n /= LT

586622 #-}

587623

588624-- -----------------------------------------------------------------------------

@@ -598,7 +634,7 @@ map f t = unstream (S.map (safe . f) (stream t))

598634-- 'Text's and concatenates the list after interspersing the first

599635-- argument between each element of the list.

600636intercalate :: Text -> [Text] -> Text

601−intercalate t = concat . (U.intersperse t)

637+intercalate t = concat . (F.intersperse t)

602638{-# INLINE intercalate #-}

603639

604640-- | /O(n)/ The 'intersperse' function takes a character and places it

@@ -613,33 +649,51 @@ reverse :: Text -> Text

613649reverse t = S.reverse (stream t)

614650{-# INLINE reverse #-}

615651

616−-- | /O(m+n)/ Replace every occurrence of one substring with another.

652+-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in

653+-- @haystack@ with @replacement@.

654+--

655+-- This function behaves as though it was defined as follows:

656+--

657+-- @

658+-- replace needle replacement haystack =

659+-- 'intercalate' replacement ('splitOn' needle haystack)

660+-- @

661+--

662+-- As this suggests, each occurrence is replaced exactly once. So if

663+-- @needle@ occurs in @replacement@, that occurrence will /not/ itself

664+-- be replaced recursively:

665+--

666+-- > replace "oo" "foo" "oo" == "foo"

667+--

668+-- In cases where several instances of @needle@ overlap, only the

669+-- first one will be replaced:

670+--

671+-- > replace "ofo" "bar" "ofofo" == "barfo"

617672--

618673-- In (unlikely) bad cases, this function's time complexity degrades

619674-- towards /O(n*m)/.

620−replace :: Text -- ^ Text to search for

621− -> Text -- ^ Replacement text

622− -> Text -- ^ Input text

675+replace :: Text

676+ -- ^ @needle@ to search for. If this string is empty, an

677+ -- error will occur.

678+ -> Text

679+ -- ^ @replacement@ to replace @needle@ with.

623680 -> Text

624−replace s d = intercalate d . splitOn s

625−{-# INLINE replace #-}

626−

627−replace2 :: Text -> Text -> Text -> Text

628−replace2 needle@(Text _ _ neeLen)

629− r@(Text repArr repOff repLen)

630− haystack@(Text hayArr hayOff hayLen)

631− | neeLen == 0 = emptyError "replace2"

632− | len < 0 = overflowError "replace2"

633− | len == 0 = empty

681+ -- ^ @haystack@ in which to search.

682+ -> Text

683+replace needle@(Text _ _ neeLen)

684+ (Text repArr repOff repLen)

685+ haystack@(Text hayArr hayOff hayLen)

686+ | neeLen == 0 = emptyError "replace"

634687 | L.null ixs = haystack

635− | otherwise = Text (A.run x) 0 len

688+ | len > 0 = Text (A.run x) 0 len

689+ | otherwise = empty

636690 where

637691 ixs = indices needle haystack

638− cnt = L.length ixs

639− len = hayLen - (neeLen - repLen) * cnt

692+ len = hayLen - (neeLen - repLen) `mul` L.length ixs

693+ x :: ST s (A.MArray s)

640694 x = do

641695 marr <- A.new len

642− let loop is0@(i:is) o d = do

696+ let loop (i:is) o d = do

643697 let d0 = d + i - o

644698 d1 = d0 + repLen

645699 A.copyI marr d hayArr (hayOff+o) d0

@@ -666,9 +720,8 @@ replace2 needle@(Text _ _ neeLen)

666720-- /Note/: In some languages, case conversion is a locale- and

667721-- context-dependent operation. The case conversion functions in this

668722-- module are /not/ locale sensitive. Programs that require locale

669−-- sensitivity should use appropriate versions of the case mapping

670−-- functions from the @text-icu@ package:

671−-- <http://hackage.haskell.org/package/text-icu>

723+-- sensitivity should use appropriate versions of the

724+-- <http://hackage.haskell.org/package/text-icu-0.6.3.7/docs/Data-Text-ICU.html#g:4 case mapping functions from the text-icu package >.

672725

673726-- | /O(n)/ Convert a string to folded case. Subject to fusion.

674727--

@@ -688,7 +741,7 @@ replace2 needle@(Text _ _ neeLen)

688741-- instead of itself.

689742toCaseFold :: Text -> Text

690743toCaseFold t = unstream (S.toCaseFold (stream t))

691−{-# INLINE [0] toCaseFold #-}

744+{-# INLINE toCaseFold #-}

692745

693746-- | /O(n)/ Convert a string to lower case, using simple case

694747-- conversion. Subject to fusion.

@@ -848,6 +901,7 @@ concat ts = case ts' of

848901 where

849902 ts' = L.filter (not . null) ts

850903 len = sumP "concat" $ L.map lengthWord16 ts'

904+ go :: ST s (A.MArray s)

851905 go = do

852906 arr <- A.new len

853907 let step i (Text a o l) =

@@ -861,13 +915,13 @@ concatMap f = concat . foldr ((:) . f) []

861915{-# INLINE concatMap #-}

862916

863917-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the

864−-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.

918+-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.

865919any :: (Char -> Bool) -> Text -> Bool

866920any p t = S.any p (stream t)

867921{-# INLINE any #-}

868922

869923-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the

870−-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.

924+-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.

871925all :: (Char -> Bool) -> Text -> Bool

872926all p t = S.all p (stream t)

873927{-# INLINE all #-}

@@ -958,10 +1012,10 @@ replicate n t@(Text a o l)

9581012 | n <= 0 || l <= 0 = empty

9591013 | n == 1 = t

9601014 | isSingleton t = replicateChar n (unsafeHead t)

961− | n <= maxBound `quot` l = Text (A.run x) 0 len

962− | otherwise = overflowError "replicate"

1015+ | otherwise = Text (A.run x) 0 len

9631016 where

964− len = l * n

1017+ len = l `mul` n

1018+ x :: ST s (A.MArray s)

9651019 x = do

9661020 arr <- A.new len

9671021 let loop !d !i | i >= n = return arr

@@ -1012,14 +1066,16 @@ take :: Int -> Text -> Text

10121066take n t@(Text arr off len)

10131067 | n <= 0 = empty

10141068 | n >= len = t

1015− | otherwise = Text arr off (loop 0 0)

1016− where

1017− loop !i !cnt

1018− | i >= len || cnt >= n = i

1019− | otherwise = loop (i+d) (cnt+1)

1020− where d = iter_ t i

1069+ | otherwise = text arr off (iterN n t)

10211070{-# INLINE [1] take #-}

10221071

1072+iterN :: Int -> Text -> Int

1073+iterN n t@(Text _arr _off len) = loop 0 0

1074+ where loop !i !cnt

1075+ | i >= len || cnt >= n = i

1076+ | otherwise = loop (i+d) (cnt+1)

1077+ where d = iter_ t i

1078+

10231079{-# RULES

10241080"TEXT take -> fused" [~1] forall n t.

10251081 take n t = unstream (S.take n (stream t))

@@ -1027,6 +1083,27 @@ take n t@(Text arr off len)

10271083 unstream (S.take n (stream t)) = take n t

10281084 #-}

10291085

1086+-- | /O(n)/ 'takeEnd' @n@ @t@ returns the suffix remaining after

1087+-- taking @n@ characters from the end of @t@.

1088+--

1089+-- Examples:

1090+--

1091+-- > takeEnd 3 "foobar" == "bar"

1092+takeEnd :: Int -> Text -> Text

1093+takeEnd n t@(Text arr off len)

1094+ | n <= 0 = empty

1095+ | n >= len = t

1096+ | otherwise = text arr (off+i) (len-i)

1097+ where i = iterNEnd n t

1098+

1099+iterNEnd :: Int -> Text -> Int

1100+iterNEnd n t@(Text _arr _off len) = loop (len-1) n

1101+ where loop i !m

1102+ | i <= 0 = 0

1103+ | m <= 1 = i

1104+ | otherwise = loop (i+d) (m-1)

1105+ where d = reverseIter_ t i

1106+

10301107-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the

10311108-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@

10321109-- is greater than the length of the 'Text'. Subject to fusion.

@@ -1034,11 +1111,8 @@ drop :: Int -> Text -> Text

10341111drop n t@(Text arr off len)

10351112 | n <= 0 = t

10361113 | n >= len = empty

1037− | otherwise = loop 0 0

1038− where loop !i !cnt

1039− | i >= len || cnt >= n = Text arr (off+i) (len-i)

1040− | otherwise = loop (i+d) (cnt+1)

1041− where d = iter_ t i

1114+ | otherwise = text arr (off+i) (len-i)

1115+ where i = iterN n t

10421116{-# INLINE [1] drop #-}

10431117

10441118{-# RULES

@@ -1048,6 +1122,18 @@ drop n t@(Text arr off len)

10481122 unstream (S.drop n (stream t)) = drop n t

10491123 #-}

10501124

1125+-- | /O(n)/ 'dropEnd' @n@ @t@ returns the prefix remaining after

1126+-- dropping @n@ characters from the end of @t@.

1127+--

1128+-- Examples:

1129+--

1130+-- > dropEnd 3 "foobar" == "foo"

1131+dropEnd :: Int -> Text -> Text

1132+dropEnd n t@(Text arr off len)

1133+ | n <= 0 = t

1134+ | n >= len = empty

1135+ | otherwise = text arr off (iterNEnd n t)

1136+

10511137-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',

10521138-- returns the longest prefix (possibly empty) of elements that

10531139-- satisfy @p@. Subject to fusion.

@@ -1055,7 +1141,7 @@ takeWhile :: (Char -> Bool) -> Text -> Text

10551141takeWhile p t@(Text arr off len) = loop 0

10561142 where loop !i | i >= len = t

10571143 | p c = loop (i+d)

1058− | otherwise = textP arr off i

1144+ | otherwise = text arr off i

10591145 where Iter c d = iter t i

10601146{-# INLINE [1] takeWhile #-}

10611147

@@ -1066,6 +1152,27 @@ takeWhile p t@(Text arr off len) = loop 0

10661152 unstream (S.takeWhile p (stream t)) = takeWhile p t

10671153 #-}

10681154

1155+-- | /O(n)/ 'takeWhileEnd', applied to a predicate @p@ and a 'Text',

1156+-- returns the longest suffix (possibly empty) of elements that

1157+-- satisfy @p@. Subject to fusion.

1158+-- Examples:

1159+--

1160+-- > takeWhileEnd (=='o') "foo" == "oo"

1161+takeWhileEnd :: (Char -> Bool) -> Text -> Text

1162+takeWhileEnd p t@(Text arr off len) = loop (len-1) len

1163+ where loop !i !l | l <= 0 = t

1164+ | p c = loop (i+d) (l+d)

1165+ | otherwise = text arr (off+l) (len-l)

1166+ where (c,d) = reverseIter t i

1167+{-# INLINE [1] takeWhileEnd #-}

1168+

1169+{-# RULES

1170+"TEXT takeWhileEnd -> fused" [~1] forall p t.

1171+ takeWhileEnd p t = S.reverse (S.takeWhile p (S.reverseStream t))

1172+"TEXT takeWhileEnd -> unfused" [1] forall p t.

1173+ S.reverse (S.takeWhile p (S.reverseStream t)) = takeWhileEnd p t

1174+ #-}

1175+

10691176-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

10701177-- 'takeWhile' @p@ @t@. Subject to fusion.

10711178dropWhile :: (Char -> Bool) -> Text -> Text

@@ -1084,8 +1191,9 @@ dropWhile p t@(Text arr off len) = loop 0 0

10841191 #-}

10851192

10861193-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

1087−-- dropping characters that fail the predicate @p@ from the end of

1194+-- dropping characters that satisfy the predicate @p@ from the end of

10881195-- @t@. Subject to fusion.

1196+--

10891197-- Examples:

10901198--

10911199-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1105,7 +1213,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len

11051213 #-}

11061214

11071215-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

1108−-- dropping characters that fail the predicate @p@ from both the

1216+-- dropping characters that satisfy the predicate @p@ from both the

11091217-- beginning and end of @t@. Subject to fusion.

11101218dropAround :: (Char -> Bool) -> Text -> Text

11111219dropAround p = dropWhile p . dropWhileEnd p

@@ -1140,13 +1248,8 @@ splitAt :: Int -> Text -> (Text, Text)

11401248splitAt n t@(Text arr off len)

11411249 | n <= 0 = (empty, t)

11421250 | n >= len = (t, empty)

1143− | otherwise = (Text arr off k, Text arr (off+k) (len-k))

1144− where k = loop 0 0

1145− loop !i !cnt

1146− | i >= len || cnt >= n = i

1147− | otherwise = loop (i+d) (cnt+1)

1148− where d = iter_ t i

1149−{-# INLINE splitAt #-}

1251+ | otherwise = let k = iterN n t

1252+ in (text arr off k, text arr (off+k) (len-k))

11501253

11511254-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns

11521255-- a pair whose first element is the longest prefix (possibly empty)

@@ -1205,9 +1308,9 @@ tails t | null t = [empty]

12051308-- copies to create substrings; they just construct new 'Text's that

12061309-- are slices of the original.

12071310

1208−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first

1209−-- 'Text' argument, consuming the delimiter. An empty delimiter is

1210−-- invalid, and will cause an error to be raised.

1311+-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'

1312+-- argument (which cannot be empty), consuming the delimiter. An empty

1313+-- delimiter is invalid, and will cause an error to be raised.

12111314--

12121315-- Examples:

12131316--

@@ -1220,16 +1323,23 @@ tails t | null t = [empty]

12201323-- > intercalate s . splitOn s == id

12211324-- > splitOn (singleton c) == split (==c)

12221325--

1326+-- (Note: the string @s@ to split on above cannot be empty.)

1327+--

12231328-- In (unlikely) bad cases, this function's time complexity degrades

12241329-- towards /O(n*m)/.

1225−splitOn :: Text -> Text -> [Text]

1330+splitOn :: Text

1331+ -- ^ String to split on. If this string is empty, an error

1332+ -- will occur.

1333+ -> Text

1334+ -- ^ Input text.

1335+ -> [Text]

12261336splitOn pat@(Text _ _ l) src@(Text arr off len)

12271337 | l <= 0 = emptyError "splitOn"

12281338 | isSingleton pat = split (== unsafeHead pat) src

12291339 | otherwise = go 0 (indices pat src)

12301340 where

1231− go !s (x:xs) = textP arr (s+off) (x-s) : go (x+l) xs

1232− go s _ = [textP arr (s+off) (len-s)]

1341+ go !s (x:xs) = text arr (s+off) (x-s) : go (x+l) xs

1342+ go s _ = [text arr (s+off) (len-s)]

12331343{-# INLINE [1] splitOn #-}

12341344

12351345{-# RULES

@@ -1321,7 +1431,7 @@ breakOn pat src@(Text arr off len)

13211431 | null pat = emptyError "breakOn"

13221432 | otherwise = case indices pat src of

13231433 [] -> (src, empty)

1324− (x:_) -> (textP arr off x, textP arr (off+x) (len-x))

1434+ (x:_) -> (text arr off x, text arr (off+x) (len-x))

13251435{-# INLINE breakOn #-}

13261436

13271437-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the

@@ -1363,7 +1473,7 @@ breakOnAll pat src@(Text arr off slen)

13631473 | otherwise = L.map step (indices pat src)

13641474 where

13651475 step x = (chunk 0 x, chunk x (slen-x))

1366− chunk !n !l = textP arr (n+off) l

1476+ chunk !n !l = text arr (n+off) l

13671477{-# INLINE breakOnAll #-}

13681478

13691479-------------------------------------------------------------------------------

@@ -1430,7 +1540,7 @@ countChar c t = S.countChar c (stream t)

14301540-- equivalent to a pair of 'unpack' operations.

14311541zip :: Text -> Text -> [(Char,Char)]

14321542zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)

1433−{-# INLINE [0] zip #-}

1543+{-# INLINE zip #-}

14341544

14351545-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function

14361546-- given as the first argument, instead of a tupling function.

@@ -1438,7 +1548,7 @@ zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)

14381548zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text

14391549zipWith f t1 t2 = unstream (S.zipWith g (stream t1) (stream t2))

14401550 where g a b = safe (f a b)

1441−{-# INLINE [0] zipWith #-}

1551+{-# INLINE zipWith #-}

14421552

14431553-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's

14441554-- representing white space.

@@ -1563,7 +1673,7 @@ isInfixOf needle haystack

15631673-- > fnordLength _ = -1

15641674stripPrefix :: Text -> Text -> Maybe Text

15651675stripPrefix p@(Text _arr _off plen) t@(Text arr off len)

1566− | p `isPrefixOf` t = Just $! textP arr (off+plen) (len-plen)

1676+ | p `isPrefixOf` t = Just $! text arr (off+plen) (len-plen)

15671677 | otherwise = Nothing

15681678

15691679-- | /O(n)/ Find the longest non-empty common prefix of two strings

@@ -1583,8 +1693,8 @@ commonPrefixes t0@(Text arr0 off0 len0) t1@(Text arr1 off1 len1) = go 0 0

15831693 where

15841694 go !i !j | i < len0 && j < len1 && a == b = go (i+d0) (j+d1)

15851695 | i > 0 = Just (Text arr0 off0 i,

1586− textP arr0 (off0+i) (len0-i),

1587− textP arr1 (off1+j) (len1-j))

1696+ text arr0 (off0+i) (len0-i),

1697+ text arr1 (off1+j) (len1-j))

15881698 | otherwise = Nothing

15891699 where Iter a d0 = iter t0 i

15901700 Iter b d1 = iter t1 j

@@ -1609,7 +1719,7 @@ commonPrefixes t0@(Text arr0 off0 len0) t1@(Text arr1 off1 len1) = go 0 0

16091719-- > quuxLength _ = -1

16101720stripSuffix :: Text -> Text -> Maybe Text

16111721stripSuffix p@(Text _arr _off plen) t@(Text arr off len)

1612− | p `isSuffixOf` t = Just $! textP arr off (len-plen)

1722+ | p `isSuffixOf` t = Just $! text arr off (len-plen)

16131723 | otherwise = Nothing

16141724

16151725-- | Add a list of non-negative numbers. Errors out on overflow.

@@ -1638,6 +1748,7 @@ overflowError fun = P.error $ "Data.Text." ++ fun ++ ": size overflow"

16381748copy :: Text -> Text

16391749copy (Text arr off len) = Text (A.run go) 0 len

16401750 where

1751+ go :: ST s (A.MArray s)

16411752 go = do

16421753 marr <- A.new len

16431754 A.copyI marr 0 arr off len

Data/Text/Array.hs+10−10

@@ -6,8 +6,7 @@

77--

88-- License : BSD-style

9−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

10−-- duncan@haskell.org

9+-- Maintainer : bos@serpentine.com

1110-- Stability : experimental

1211-- Portability : portable

1312--

@@ -21,13 +20,13 @@

2120-- > import qualified Data.Text.Array as A

2221--

2322-- The names in this module resemble those in the 'Data.Array' family

24−-- of modules, but are shorter due to the assumption of qualifid

23+-- of modules, but are shorter due to the assumption of qualified

2524-- naming.

2625module Data.Text.Array

2726 (

2827 -- * Types

29− Array(aBA)

30− , MArray(maBA)

28+ Array(Array, aBA)

29+ , MArray(MArray, maBA)

3130

3231 -- * Functions

3332 , copyM

@@ -66,8 +65,8 @@ import Control.Monad.ST.Unsafe (unsafeIOToST)

6665import Control.Monad.ST (unsafeIOToST)

6766#endif

6867import Data.Bits ((.&.), xor)

69−import Data.Text.Unsafe.Base (inlinePerformIO)

70−import Data.Text.UnsafeShift (shiftL, shiftR)

68+import Data.Text.Internal.Unsafe (inlinePerformIO)

69+import Data.Text.Internal.Unsafe.Shift (shiftL, shiftR)

7170#if __GLASGOW_HASKELL__ >= 703

7271import Foreign.C.Types (CInt(CInt), CSize(CSize))

7372#else

@@ -75,7 +74,7 @@ import Foreign.C.Types (CInt, CSize)

7574#endif

7675import GHC.Base (ByteArray#, MutableByteArray#, Int(..),

7776 indexWord16Array#, newByteArray#,

78− unsafeCoerce#, writeWord16Array#)

77+ unsafeFreezeByteArray#, writeWord16Array#)

7978import GHC.ST (ST(..), runST)

8079import GHC.Word (Word16(..))

8180import Prelude hiding (length, read)

@@ -131,8 +130,9 @@ array_size_error = error "Data.Text.Array.new: size overflow"

131130

132131-- | Freeze a mutable array. Do not mutate the 'MArray' afterwards!

133132unsafeFreeze :: MArray s -> ST s Array

134−unsafeFreeze MArray{..} = ST $ \s# ->

135− (# s#, Array (unsafeCoerce# maBA)

133+unsafeFreeze MArray{..} = ST $ \s1# ->

134+ case unsafeFreezeByteArray# maBA s1# of

135+ (# s2#, ba# #) -> (# s2#, Array ba#

136136#if defined(ASSERTS)

137137 maLen

138138#endif

Data/Text/Encoding.hs+135−91

@@ -10,8 +10,7 @@

1111--

1212-- License : BSD-style

13−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

14−-- duncan@haskell.org

13+-- Maintainer : bos@serpentine.com

1514-- Stability : experimental

1615-- Portability : portable

1716--

@@ -55,36 +54,45 @@ module Data.Text.Encoding

5554 , encodeUtf16BE

5655 , encodeUtf32LE

5756 , encodeUtf32BE

57+

58+ -- * Encoding Text using ByteString Builders

59+ , encodeUtf8Builder

60+ , encodeUtf8BuilderEscaped

5861 ) where

5962

60−import Control.Exception (evaluate, try)

6163#if __GLASGOW_HASKELL__ >= 702

6264import Control.Monad.ST.Unsafe (unsafeIOToST, unsafeSTToIO)

6365#else

6466import Control.Monad.ST (unsafeIOToST, unsafeSTToIO)

6567#endif

68+

69+import Control.Exception (evaluate, try)

6670import Control.Monad.ST (runST)

6771import Data.Bits ((.&.))

6872import Data.ByteString as B

69−import Data.ByteString.Internal as B

70−import Data.Text ()

73+import Data.ByteString.Internal as B hiding (c2w)

7174import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

72−import Data.Text.Internal (Text(..), safe, textP)

73−import Data.Text.Private (runText)

74−import Data.Text.UnsafeChar (ord, unsafeWrite)

75−import Data.Text.UnsafeShift (shiftL, shiftR)

75+import Data.Text.Internal (Text(..), safe, text)

76+import Data.Text.Internal.Private (runText)

77+import Data.Text.Internal.Unsafe.Char (ord, unsafeWrite)

78+import Data.Text.Internal.Unsafe.Shift (shiftR)

79+import Data.Text.Show ()

80+import Data.Text.Unsafe (unsafeDupablePerformIO)

7681import Data.Word (Word8, Word32)

77−import Foreign.C.Types (CSize)

82+import Foreign.C.Types (CSize(..))

7883import Foreign.ForeignPtr (withForeignPtr)

7984import Foreign.Marshal.Utils (with)

8085import Foreign.Ptr (Ptr, minusPtr, nullPtr, plusPtr)

8186import Foreign.Storable (Storable, peek, poke)

82−import GHC.Base (MutableByteArray#)

87+import GHC.Base (ByteArray#, MutableByteArray#)

88+import qualified Data.ByteString.Builder as B

89+import qualified Data.ByteString.Builder.Internal as B hiding (empty, append)

90+import qualified Data.ByteString.Builder.Prim as BP

91+import qualified Data.ByteString.Builder.Prim.Internal as BP

8392import qualified Data.Text.Array as A

84−import qualified Data.Text.Encoding.Fusion as E

85−import qualified Data.Text.Encoding.Utf16 as U16

86−import qualified Data.Text.Fusion as F

87−import Data.Text.Unsafe (unsafeDupablePerformIO)

93+import qualified Data.Text.Internal.Encoding.Fusion as E

94+import qualified Data.Text.Internal.Encoding.Utf16 as U16

95+import qualified Data.Text.Internal.Fusion as F

8896

8997#include "text_cbits.h"

9098

@@ -103,8 +111,6 @@ import Data.Text.Unsafe (unsafeDupablePerformIO)

103111

104112-- | /Deprecated/. Decode a 'ByteString' containing 7-bit ASCII

105113-- encoded text.

106−--

107−-- This function is deprecated. Use 'decodeLatin1' instead.

108114decodeASCII :: ByteString -> Text

109115decodeASCII = decodeUtf8

110116{-# DEPRECATED decodeASCII "Use decodeUtf8 instead" #-}

@@ -114,7 +120,7 @@ decodeASCII = decodeUtf8

114120-- 'decodeLatin1' is semantically equivalent to

115121-- @Data.Text.pack . Data.ByteString.Char8.unpack@

116122decodeLatin1 :: ByteString -> Text

117−decodeLatin1 (PS fp off len) = textP a 0 len

123+decodeLatin1 (PS fp off len) = text a 0 len

118124 where

119125 a = A.run (A.new len >>= unsafeIOToST . go)

120126 go dest = withForeignPtr fp $ \ptr -> do

@@ -146,7 +152,7 @@ decodeUtf8With onErr (PS fp off len) = runText $ \done -> do

146152 loop (ptr `plusPtr` off)

147153 (unsafeIOToST . go) =<< A.new len

148154 where

149− desc = "Data.Text.Encoding.decodeUtf8: Invalid UTF-8 stream"

155+ desc = "Data.Text.Internal.Encoding.decodeUtf8: Invalid UTF-8 stream"

150156{- INLINE[0] decodeUtf8With #-}

151157

152158-- $stream

@@ -155,11 +161,15 @@ decodeUtf8With onErr (PS fp off len) = runText $ \done -> do

155161-- a 'ByteString' that represents a possibly incomplete input (e.g. a

156162-- packet from a network stream) that may not end on a UTF-8 boundary.

157163--

158−-- The first element of the result is the maximal chunk of 'Text' that

159−-- can be decoded from the given input. The second is a function which

160−-- accepts another 'ByteString'. That string will be assumed to

161−-- directly follow the string that was passed as input to the original

162−-- function, and it will in turn be decoded.

164+-- 1. The maximal prefix of 'Text' that could be decoded from the

165+-- given input.

166+--

167+-- 2. The suffix of the 'ByteString' that could not be decoded due to

168+-- insufficient input.

169+--

170+-- 3. A function that accepts another 'ByteString'. That string will

171+-- be assumed to directly follow the string that was passed as

172+-- input to the original function, and it will in turn be decoded.

163173--

164174-- To help understand the use of these functions, consider the Unicode

165175-- string @\"hi ☃\"@. If encoded as UTF-8, this becomes @\"hi

@@ -172,15 +182,17 @@ decodeUtf8With onErr (PS fp off len) = runText $ \done -> do

172182-- as we receive each one.

173183--

174184-- @

175−-- let 'Some' t0 f0 = 'streamDecodeUtf8' \"hi \\xe2\"

176−-- t0 == \"hi \" :: 'Text'

185+-- ghci> let s0\@('Some' _ _ f0) = 'streamDecodeUtf8' \"hi \\xe2\"

186+-- ghci> s0

187+-- 'Some' \"hi \" \"\\xe2\" _

177188-- @

178189--

179190-- We use the continuation @f0@ to decode our second packet.

180191--

181192-- @

182−-- let 'Some' t1 f1 = f0 \"\\x98\"

183−-- t1 == \"\"

193+-- ghci> let s1\@('Some' _ _ f1) = f0 \"\\x98\"

194+-- ghci> s1

195+-- 'Some' \"\" \"\\xe2\\x98\"

184196-- @

185197--

186198-- We could not give @f0@ enough input to decode anything, so it

@@ -188,8 +200,9 @@ decodeUtf8With onErr (PS fp off len) = runText $ \done -> do

188200-- the last byte of input, it will make progress.

189201--

190202-- @

191−-- let 'Some' t2 f2 = f1 \"\\x83\"

192−-- t2 == \"☃\"

203+-- ghci> let s2\@('Some' _ _ f2) = f1 \"\\x83\"

204+-- ghci> s2

205+-- 'Some' \"\\x2603\" \"\" _

193206-- @

194207--

195208-- If given invalid input, an exception will be thrown by the function

@@ -221,12 +234,13 @@ streamDecodeUtf8 = streamDecodeUtf8With strictDecode

221234-- | Decode, in a stream oriented way, a 'ByteString' containing UTF-8

222235-- encoded text.

223236streamDecodeUtf8With :: OnDecodeError -> ByteString -> Decoding

224−streamDecodeUtf8With onErr = decodeChunk 0 0

237+streamDecodeUtf8With onErr = decodeChunk B.empty 0 0

225238 where

226239 -- We create a slightly larger than necessary buffer to accommodate a

227240 -- potential surrogate pair started in the last buffer

228− decodeChunk :: CodePoint -> DecoderState -> ByteString -> Decoding

229− decodeChunk codepoint0 state0 bs@(PS fp off len) =

241+ decodeChunk :: ByteString -> CodePoint -> DecoderState -> ByteString

242+ -> Decoding

243+ decodeChunk undecoded0 codepoint0 state0 bs@(PS fp off len) =

230244 runST $ (unsafeIOToST . decodeChunkToBuffer) =<< A.new (len+1)

231245 where

232246 decodeChunkToBuffer :: A.MArray s -> IO Decoding

@@ -245,6 +259,7 @@ streamDecodeUtf8With onErr = decodeChunk 0 0

245259 UTF8_REJECT -> do

246260 -- We encountered an encoding error

247261 x <- peek curPtr'

262+ poke statePtr 0

248263 case onErr desc (Just x) of

249264 Nothing -> loop $ curPtr' `plusPtr` 1

250265 Just c -> do

@@ -252,7 +267,6 @@ streamDecodeUtf8With onErr = decodeChunk 0 0

252267 w <- unsafeSTToIO $

253268 unsafeWrite dest (fromIntegral destOff) (safe c)

254269 poke destOffPtr (destOff + fromIntegral w)

255− poke statePtr 0

256270 loop $ curPtr' `plusPtr` 1

257271

258272 _ -> do

@@ -261,13 +275,16 @@ streamDecodeUtf8With onErr = decodeChunk 0 0

261275 codepoint <- peek codepointPtr

262276 chunkText <- unsafeSTToIO $ do

263277 arr <- A.unsafeFreeze dest

264− return $! textP arr 0 (fromIntegral n)

278+ return $! text arr 0 (fromIntegral n)

265279 lastPtr <- peek curPtrPtr

266280 let left = lastPtr `minusPtr` curPtr

267− return $ Some chunkText (B.drop left bs)

268− (decodeChunk codepoint state)

281+ !undecoded = case state of

282+ UTF8_ACCEPT -> B.empty

283+ _ -> B.append undecoded0 (B.drop left bs)

284+ return $ Some chunkText undecoded

285+ (decodeChunk undecoded codepoint state)

269286 in loop (ptr `plusPtr` off)

270− desc = "Data.Text.Encoding.streamDecodeUtf8With: Invalid UTF-8 stream"

287+ desc = "Data.Text.Internal.Encoding.streamDecodeUtf8With: Invalid UTF-8 stream"

271288

272289-- | Decode a 'ByteString' containing UTF-8 encoded text that is known

273290-- to be valid.

@@ -290,61 +307,85 @@ decodeUtf8' :: ByteString -> Either UnicodeException Text

290307decodeUtf8' = unsafeDupablePerformIO . try . evaluate . decodeUtf8With strictDecode

291308{-# INLINE decodeUtf8' #-}

292309

310+-- | Encode text to a ByteString 'B.Builder' using UTF-8 encoding.

311+encodeUtf8Builder :: Text -> B.Builder

312+encodeUtf8Builder = encodeUtf8BuilderEscaped (BP.liftFixedToBounded BP.word8)

313+

314+-- | Encode text using UTF-8 encoding and escape the ASCII characters using

315+-- a 'BP.BoundedPrim'.

316+--

317+-- Use this function is to implement efficient encoders for text-based formats

318+-- like JSON or HTML.

319+{-# INLINE encodeUtf8BuilderEscaped #-}

320+-- TODO: Extend documentation with references to source code in @blaze-html@

321+-- or @aeson@ that uses this function.

322+encodeUtf8BuilderEscaped :: BP.BoundedPrim Word8 -> Text -> B.Builder

323+encodeUtf8BuilderEscaped be =

324+ -- manual eta-expansion to ensure inlining works as expected

325+ \txt -> B.builder (mkBuildstep txt)

326+ where

327+ bound = max 4 $ BP.sizeBound be

328+

329+ mkBuildstep (Text arr off len) !k =

330+ outerLoop off

331+ where

332+ iend = off + len

333+

334+ outerLoop !i0 !br@(B.BufferRange op0 ope)

335+ | i0 >= iend = k br

336+ | outRemaining > 0 = goPartial (i0 + min outRemaining inpRemaining)

337+ -- TODO: Use a loop with an integrated bound's check if outRemaining

338+ -- is smaller than 8, as this will save on divisions.

339+ | otherwise = return $ B.bufferFull bound op0 (outerLoop i0)

340+ where

341+ outRemaining = (ope `minusPtr` op0) `div` bound

342+ inpRemaining = iend - i0

343+

344+ goPartial !iendTmp = go i0 op0

345+ where

346+ go !i !op

347+ | i < iendTmp = case A.unsafeIndex arr i of

348+ w | w <= 0x7F -> do

349+ BP.runB be (fromIntegral w) op >>= go (i + 1)

350+ | w <= 0x7FF -> do

351+ poke8 0 $ (w `shiftR` 6) + 0xC0

352+ poke8 1 $ (w .&. 0x3f) + 0x80

353+ go (i + 1) (op `plusPtr` 2)

354+ | 0xD800 <= w && w <= 0xDBFF -> do

355+ let c = ord $ U16.chr2 w (A.unsafeIndex arr (i+1))

356+ poke8 0 $ (c `shiftR` 18) + 0xF0

357+ poke8 1 $ ((c `shiftR` 12) .&. 0x3F) + 0x80

358+ poke8 2 $ ((c `shiftR` 6) .&. 0x3F) + 0x80

359+ poke8 3 $ (c .&. 0x3F) + 0x80

360+ go (i + 2) (op `plusPtr` 4)

361+ | otherwise -> do

362+ poke8 0 $ (w `shiftR` 12) + 0xE0

363+ poke8 1 $ ((w `shiftR` 6) .&. 0x3F) + 0x80

364+ poke8 2 $ (w .&. 0x3F) + 0x80

365+ go (i + 1) (op `plusPtr` 3)

366+ | otherwise =

367+ outerLoop i (B.BufferRange op ope)

368+ where

369+ poke8 j v = poke (op `plusPtr` j) (fromIntegral v :: Word8)

370+

293371-- | Encode text using UTF-8 encoding.

294372encodeUtf8 :: Text -> ByteString

295−encodeUtf8 (Text arr off len) = unsafeDupablePerformIO $ do

296− let size0 = max len 4

297− mallocByteString size0 >>= start size0 off 0

298− where

299− start size n0 m0 fp = withForeignPtr fp $ loop n0 m0

300− where

301− loop n1 m1 ptr = go n1 m1

302− where

303− offLen = off + len

304− go !n !m

305− | n == offLen = return (PS fp 0 m)

306− | otherwise = do

307− let poke8 k v = poke (ptr `plusPtr` k) (fromIntegral v :: Word8)

308− ensure k act

309− | size-m >= k = act

310− | otherwise = {-# SCC "resizeUtf8/ensure" #-} do

311− let newSize = size `shiftL` 1

312− fp' <- mallocByteString newSize

313− withForeignPtr fp' $ \ptr' ->

314− memcpy ptr' ptr (fromIntegral m)

315− start newSize n m fp'

316− {-# INLINE ensure #-}

317− case A.unsafeIndex arr n of

318− w| w <= 0x7F -> ensure 1 $ do

319− poke (ptr `plusPtr` m) (fromIntegral w :: Word8)

320− -- A single ASCII octet is likely to start a run of

321− -- them. We see better performance when we

322− -- special-case this assumption.

323− let end = ptr `plusPtr` size

324− ascii !t !u

325− | t == offLen || u == end || v >= 0x80 =

326− go t (u `minusPtr` ptr)

327− | otherwise = do

328− poke u (fromIntegral v :: Word8)

329− ascii (t+1) (u `plusPtr` 1)

330− where v = A.unsafeIndex arr t

331− ascii (n+1) (ptr `plusPtr` (m+1))

332− | w <= 0x7FF -> ensure 2 $ do

333− poke8 m $ (w `shiftR` 6) + 0xC0

334− poke8 (m+1) $ (w .&. 0x3f) + 0x80

335− go (n+1) (m+2)

336− | 0xD800 <= w && w <= 0xDBFF -> ensure 4 $ do

337− let c = ord $ U16.chr2 w (A.unsafeIndex arr (n+1))

338− poke8 m $ (c `shiftR` 18) + 0xF0

339− poke8 (m+1) $ ((c `shiftR` 12) .&. 0x3F) + 0x80

340− poke8 (m+2) $ ((c `shiftR` 6) .&. 0x3F) + 0x80

341− poke8 (m+3) $ (c .&. 0x3F) + 0x80

342− go (n+2) (m+4)

343− | otherwise -> ensure 3 $ do

344− poke8 m $ (w `shiftR` 12) + 0xE0

345− poke8 (m+1) $ ((w `shiftR` 6) .&. 0x3F) + 0x80

346− poke8 (m+2) $ (w .&. 0x3F) + 0x80

347− go (n+1) (m+3)

373+encodeUtf8 (Text arr off len)

374+ | len == 0 = B.empty

375+ | otherwise = unsafeDupablePerformIO $ do

376+ fp <- mallocByteString (len*4)

377+ withForeignPtr fp $ \ptr ->

378+ with ptr $ \destPtr -> do

379+ c_encode_utf8 destPtr (A.aBA arr) (fromIntegral off) (fromIntegral len)

380+ newDest <- peek destPtr

381+ let utf8len = newDest `minusPtr` ptr

382+ if utf8len >= len `shiftR` 1

383+ then return (PS fp 0 utf8len)

384+ else do

385+ fp' <- mallocByteString utf8len

386+ withForeignPtr fp' $ \ptr' -> do

387+ memcpy ptr' ptr (fromIntegral utf8len)

388+ return (PS fp' 0 utf8len)

348389

349390-- | Decode text from little endian UTF-16 encoding.

350391decodeUtf16LEWith :: OnDecodeError -> ByteString -> Text

@@ -433,3 +474,6 @@ foreign import ccall unsafe "_hs_text_decode_utf8_state" c_decode_utf8_with_stat

433474

434475foreign import ccall unsafe "_hs_text_decode_latin1" c_decode_latin1

435476 :: MutableByteArray# s -> Ptr Word8 -> Ptr Word8 -> IO ()

477+

478+foreign import ccall unsafe "_hs_text_encode_utf8" c_encode_utf8

479+ :: Ptr (Ptr Word8) -> ByteArray# -> CSize -> CSize -> IO ()

Data/Text/Encoding/Error.hs+8−3

@@ -1,5 +1,7 @@

11{-# LANGUAGE CPP, DeriveDataTypeable #-}

2−#if __GLASGOW_HASKELL__ >= 702

2+#if __GLASGOW_HASKELL__ >= 704

3+{-# LANGUAGE Safe #-}

4+#elif __GLASGOW_HASKELL__ >= 702

35{-# LANGUAGE Trustworthy #-}

46#endif

57-- |

@@ -7,8 +9,7 @@

79-- Copyright : (c) Bryan O'Sullivan 2009

810--

911-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

12+-- Maintainer : bos@serpentine.com

1213-- Stability : experimental

1314-- Portability : GHC

1415--

@@ -67,6 +68,7 @@ type OnError a b = String -> Maybe a -> Maybe b

6768type OnDecodeError = OnError Word8 Char

6869

6970-- | A handler for an encoding error.

71+{-# DEPRECATED OnEncodeError "This exception is never used in practice, and will be removed." #-}

7072type OnEncodeError = OnError Char Word8

7173

7274-- | An exception type for representing Unicode encoding errors.

@@ -79,6 +81,8 @@ data UnicodeException =

7981 -- under the given encoding, or ran out of input in mid-encode.

8082 deriving (Eq, Typeable)

8183

84+{-# DEPRECATED EncodeError "This constructor is never used, and will be removed." #-}

85+

8286showUnicodeException :: UnicodeException -> String

8387showUnicodeException (DecodeError desc (Just w))

8488 = "Cannot decode byte '\\x" ++ showHex w ("': " ++ desc)

@@ -108,6 +112,7 @@ lenientDecode :: OnDecodeError

108112lenientDecode _ _ = Just '\xfffd'

109113

110114-- | Throw a 'UnicodeException' if encoding fails.

115+{-# DEPRECATED strictEncode "This function always throws an exception, and will be removed." #-}

111116strictEncode :: OnEncodeError

112117strictEncode desc c = throw (EncodeError desc c)

113118

Data/Text/Foreign.hs+1−2

@@ -4,8 +4,7 @@

55--

66-- License : BSD-style

7−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

8−-- duncan@haskell.org

7+-- Maintainer : bos@serpentine.com

98-- Stability : experimental

109-- Portability : GHC

1110--

Data/Text/Fusion/Size.hsdeleted−146

@@ -1,146 +0,0 @@

1−{-# LANGUAGE CPP #-}

2−{-# OPTIONS_GHC -fno-warn-missing-methods #-}

3−-- |

4−-- Module : Data.Text.Fusion.Internal

5−-- Copyright : (c) Roman Leshchinskiy 2008,

6−-- (c) Bryan O'Sullivan 2009

7−--

8−-- License : BSD-style

9−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

10−-- duncan@haskell.org

11−-- Stability : experimental

12−-- Portability : portable

13−--

14−-- Size hints.

15−

16−module Data.Text.Fusion.Size

17− (

18− Size

19− , exactly

20− , exactSize

21− , maxSize

22− , unknownSize

23− , smaller

24− , larger

25− , upperBound

26− , isEmpty

27− ) where

28−

29−#if defined(ASSERTS)

30−import Control.Exception (assert)

31−#endif

32−

33−data Size = Exact {-# UNPACK #-} !Int -- ^ Exact size.

34− | Max {-# UNPACK #-} !Int -- ^ Upper bound on size.

35− | Unknown -- ^ Unknown size.

36− deriving (Eq, Show)

37−

38−exactly :: Size -> Maybe Int

39−exactly (Exact n) = Just n

40−exactly _ = Nothing

41−{-# INLINE exactly #-}

42−

43−exactSize :: Int -> Size

44−exactSize n =

45−#if defined(ASSERTS)

46− assert (n >= 0)

47−#endif

48− Exact n

49−{-# INLINE exactSize #-}

50−

51−maxSize :: Int -> Size

52−maxSize n =

53−#if defined(ASSERTS)

54− assert (n >= 0)

55−#endif

56− Max n

57−{-# INLINE maxSize #-}

58−

59−unknownSize :: Size

60−unknownSize = Unknown

61−{-# INLINE unknownSize #-}

62−

63−instance Num Size where

64− (+) = addSize

65− (-) = subtractSize

66− (*) = mulSize

67−

68− fromInteger = f where f = Exact . fromInteger

69− {-# INLINE f #-}

70−

71−add :: Int -> Int -> Int

72−add m n | mn >= 0 = mn

73− | otherwise = overflowError

74− where mn = m + n

75−{-# INLINE add #-}

76−

77−addSize :: Size -> Size -> Size

78−addSize (Exact m) (Exact n) = Exact (add m n)

79−addSize (Exact m) (Max n) = Max (add m n)

80−addSize (Max m) (Exact n) = Max (add m n)

81−addSize (Max m) (Max n) = Max (add m n)

82−addSize _ _ = Unknown

83−{-# INLINE addSize #-}

84−

85−subtractSize :: Size -> Size -> Size

86−subtractSize (Exact m) (Exact n) = Exact (max (m-n) 0)

87−subtractSize (Exact m) (Max _) = Max m

88−subtractSize (Max m) (Exact n) = Max (max (m-n) 0)

89−subtractSize a@(Max _) (Max _) = a

90−subtractSize a@(Max _) Unknown = a

91−subtractSize _ _ = Unknown

92−{-# INLINE subtractSize #-}

93−

94−mul :: Int -> Int -> Int

95−mul m n

96− | m <= maxBound `quot` n = m * n

97− | otherwise = overflowError

98−{-# INLINE mul #-}

99−

100−mulSize :: Size -> Size -> Size

101−mulSize (Exact m) (Exact n) = Exact (mul m n)

102−mulSize (Exact m) (Max n) = Max (mul m n)

103−mulSize (Max m) (Exact n) = Max (mul m n)

104−mulSize (Max m) (Max n) = Max (mul m n)

105−mulSize _ _ = Unknown

106−{-# INLINE mulSize #-}

107−

108−-- | Minimum of two size hints.

109−smaller :: Size -> Size -> Size

110−smaller (Exact m) (Exact n) = Exact (m `min` n)

111−smaller (Exact m) (Max n) = Max (m `min` n)

112−smaller (Exact m) Unknown = Max m

113−smaller (Max m) (Exact n) = Max (m `min` n)

114−smaller (Max m) (Max n) = Max (m `min` n)

115−smaller a@(Max _) Unknown = a

116−smaller Unknown (Exact n) = Max n

117−smaller Unknown (Max n) = Max n

118−smaller Unknown Unknown = Unknown

119−{-# INLINE smaller #-}

120−

121−-- | Maximum of two size hints.

122−larger :: Size -> Size -> Size

123−larger (Exact m) (Exact n) = Exact (m `max` n)

124−larger a@(Exact m) b@(Max n) | m >= n = a

125− | otherwise = b

126−larger a@(Max m) b@(Exact n) | n >= m = b

127− | otherwise = a

128−larger (Max m) (Max n) = Max (m `max` n)

129−larger _ _ = Unknown

130−{-# INLINE larger #-}

131−

132−-- | Compute the maximum size from a size hint, if possible.

133−upperBound :: Int -> Size -> Int

134−upperBound _ (Exact n) = n

135−upperBound _ (Max n) = n

136−upperBound k _ = k

137−{-# INLINE upperBound #-}

138−

139−isEmpty :: Size -> Bool

140−isEmpty (Exact n) = n <= 0

141−isEmpty (Max n) = n <= 0

142−isEmpty _ = False

143−{-# INLINE isEmpty #-}

144−

145−overflowError :: Int

146−overflowError = error "Data.Text.Fusion.Size: size overflow"

Data/Text/IO.hs+3−3

@@ -50,9 +50,9 @@ import qualified Control.Exception as E

5050import Control.Monad (liftM2, when)

5151import Data.IORef (readIORef, writeIORef)

5252import qualified Data.Text as T

53−import Data.Text.Fusion (stream)

54−import Data.Text.Fusion.Internal (Step(..), Stream(..))

55−import Data.Text.IO.Internal (hGetLineWith, readChunk)

53+import Data.Text.Internal.Fusion (stream)

54+import Data.Text.Internal.Fusion.Types (Step(..), Stream(..))

55+import Data.Text.Internal.IO (hGetLineWith, readChunk)

5656import GHC.IO.Buffer (Buffer(..), BufferState(..), CharBufElem, CharBuffer,

5757 RawCharBuffer, emptyBuffer, isEmptyBuffer, newCharBuffer,

5858 writeCharBuf)

Data/Text/Internal.hs+77−13

@@ -1,4 +1,5 @@

1−{-# LANGUAGE CPP, DeriveDataTypeable #-}

1+{-# LANGUAGE CPP, DeriveDataTypeable, UnboxedTuples #-}

2+{-# OPTIONS_HADDOCK not-home #-}

34-- |

45-- Module : Data.Text.Internal

@@ -7,8 +8,7 @@

89--

910-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

11+-- Maintainer : bos@serpentine.com

1212-- Stability : experimental

1313-- Portability : GHC

1414--

@@ -20,6 +20,10 @@

2020-- with the internals, as the functions here do just about nothing to

2121-- preserve data invariants. You have been warned!

2222

23+#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)

24+#include "MachDeps.h"

25+#endif

26+

2327module Data.Text.Internal

2428 (

2529 -- * Types

@@ -32,8 +36,13 @@ module Data.Text.Internal

3236 , safe

3337 -- * Code that must be here for accessibility

3438 , empty

39+ , empty_

3540 -- * Utilities

3641 , firstf

42+ -- * Checked multiplication

43+ , mul

44+ , mul32

45+ , mul64

3746 -- * Debugging

3847 , showText

3948 ) where

@@ -41,10 +50,11 @@ module Data.Text.Internal

4150#if defined(ASSERTS)

4251import Control.Exception (assert)

4352#endif

44−import Data.Bits ((.&.))

45−import qualified Data.Text.Array as A

46−import Data.Text.UnsafeChar (ord)

53+import Data.Bits

54+import Data.Int (Int32, Int64)

55+import Data.Text.Internal.Unsafe.Char (ord)

4756import Data.Typeable (Typeable)

57+import qualified Data.Text.Array as A

4858

4959-- | A space efficient, packed, unboxed Unicode text type.

5060data Text = Text

@@ -54,8 +64,8 @@ data Text = Text

5464 deriving (Typeable)

5565

5666-- | Smart constructor.

57−text :: A.Array -> Int -> Int -> Text

58−text arr off len =

67+text_ :: A.Array -> Int -> Int -> Text

68+text_ arr off len =

5969#if defined(ASSERTS)

6070 let c = A.unsafeIndex arr off

6171 alen = A.length arr

@@ -65,19 +75,28 @@ text arr off len =

6575 assert (len == 0 || c < 0xDC00 || c > 0xDFFF) $

6676#endif

6777 Text arr off len

68−{-# INLINE text #-}

78+{-# INLINE text_ #-}

6979

7080-- | /O(1)/ The empty 'Text'.

7181empty :: Text

7282empty = Text A.empty 0 0

7383{-# INLINE [1] empty #-}

7484

85+-- | A non-inlined version of 'empty'.

86+empty_ :: Text

87+empty_ = Text A.empty 0 0

88+{-# NOINLINE empty_ #-}

89+

7590-- | Construct a 'Text' without invisibly pinning its byte array in

7691-- memory if its length has dwindled to zero.

92+text :: A.Array -> Int -> Int -> Text

93+text arr off len | len == 0 = empty

94+ | otherwise = text_ arr off len

95+{-# INLINE text #-}

96+

7797textP :: A.Array -> Int -> Int -> Text

78−textP arr off len | len == 0 = empty

79− | otherwise = text arr off len

80−{-# INLINE textP #-}

98+{-# DEPRECATED textP "Use text instead" #-}

99+textP = text

81100

82101-- | A useful 'show'-like function for debugging purposes.

83102showText :: Text -> String

@@ -97,13 +116,58 @@ safe :: Char -> Char

97116safe c

98117 | ord c .&. 0x1ff800 /= 0xd800 = c

99118 | otherwise = '\xfffd'

100−{-# INLINE safe #-}

119+{-# INLINE [0] safe #-}

101120

102121-- | Apply a function to the first element of an optional pair.

103122firstf :: (a -> c) -> Maybe (a,b) -> Maybe (c,b)

104123firstf f (Just (a, b)) = Just (f a, b)

105124firstf _ Nothing = Nothing

106125

126+-- | Checked multiplication. Calls 'error' if the result would

127+-- overflow.

128+mul :: Int -> Int -> Int

129+#if WORD_SIZE_IN_BITS == 64

130+mul a b = fromIntegral $ fromIntegral a `mul64` fromIntegral b

131+#else

132+mul a b = fromIntegral $ fromIntegral a `mul32` fromIntegral b

133+#endif

134+{-# INLINE mul #-}

135+infixl 7 `mul`

136+

137+-- | Checked multiplication. Calls 'error' if the result would

138+-- overflow.

139+mul64 :: Int64 -> Int64 -> Int64

140+mul64 a b

141+ | a >= 0 && b >= 0 = mul64_ a b

142+ | a >= 0 = -mul64_ a (-b)

143+ | b >= 0 = -mul64_ (-a) b

144+ | otherwise = mul64_ (-a) (-b)

145+{-# INLINE mul64 #-}

146+infixl 7 `mul64`

147+

148+mul64_ :: Int64 -> Int64 -> Int64

149+mul64_ a b

150+ | ahi > 0 && bhi > 0 = error "overflow"

151+ | top > 0x7fffffff = error "overflow"

152+ | total < 0 = error "overflow"

153+ | otherwise = total

154+ where (# ahi, alo #) = (# a `shiftR` 32, a .&. 0xffffffff #)

155+ (# bhi, blo #) = (# b `shiftR` 32, b .&. 0xffffffff #)

156+ top = ahi * blo + alo * bhi

157+ total = (top `shiftL` 32) + alo * blo

158+{-# INLINE mul64_ #-}

159+

160+-- | Checked multiplication. Calls 'error' if the result would

161+-- overflow.

162+mul32 :: Int32 -> Int32 -> Int32

163+mul32 a b = case fromIntegral a * fromIntegral b of

164+ ab | ab < min32 || ab > max32 -> error "overflow"

165+ | otherwise -> fromIntegral ab

166+ where min32 = -0x80000000 :: Int64

167+ max32 = 0x7fffffff

168+{-# INLINE mul32 #-}

169+infixl 7 `mul32`

170+

107171-- $internals

108172--

109173-- Internally, the 'Text' type is represented as an array of 'Word16'

Data/Text/Lazy/Builder/Internal.hs → Data/Text/Internal/Builder.hsrenamed+27−19

@@ -1,16 +1,21 @@

11{-# LANGUAGE BangPatterns, CPP, Rank2Types #-}

2+{-# OPTIONS_HADDOCK not-home #-}

34-----------------------------------------------------------------------------

45-- |

5−-- Module : Data.Text.Lazy.Builder.Internal

6+-- Module : Data.Text.Internal.Builder

8−-- License : BSD3-style (see LICENSE)

9+-- License : BSD-style (see LICENSE)

910--

1011-- Maintainer : Johan Tibell <johan.tibell@gmail.com>

1112-- Stability : experimental

1213-- Portability : portable to Hugs and GHC

1314--

15+-- /Warning/: this is an internal module, and does not have a stable

16+-- API or name. Functions in this module may not check or enforce

17+-- preconditions expected by public modules. Use at your own risk!

18+--

1419-- Efficient construction of lazy @Text@ values. The principal

1520-- operations on a @Builder@ are @singleton@, @fromText@, and

1621-- @fromLazyText@, which construct new builders, and 'mappend', which

@@ -30,7 +35,7 @@

3035--

3136-----------------------------------------------------------------------------

3237

33−module Data.Text.Lazy.Builder.Internal

38+module Data.Text.Internal.Builder

3439 ( -- * Public API

3540 -- ** The Builder type

3641 Builder

@@ -53,13 +58,14 @@ module Data.Text.Lazy.Builder.Internal

5358 ) where

5459

5560import Control.Monad.ST (ST, runST)

56−import Data.Bits ((.&.))

5761import Data.Monoid (Monoid(..))

62+#if MIN_VERSION_base(4,9,0)

63+import Data.Semigroup (Semigroup(..))

64+#endif

5865import Data.Text.Internal (Text(..))

59−import Data.Text.Lazy.Internal (smallChunkSize)

66+import Data.Text.Internal.Lazy (smallChunkSize)

6067import Data.Text.Unsafe (inlineInterleaveST)

61−import Data.Text.UnsafeChar (ord, unsafeWrite)

62−import Data.Text.UnsafeShift (shiftR)

68+import Data.Text.Internal.Unsafe.Char (unsafeWrite)

6369import Prelude hiding (map, putChar)

6470

6571import qualified Data.String as String

@@ -86,12 +92,22 @@ newtype Builder = Builder {

8692 -> ST s [S.Text]

8793 }

8894

95+#if MIN_VERSION_base(4,9,0)

96+instance Semigroup Builder where

97+ (<>) = append

98+ {-# INLINE (<>) #-}

99+#endif

100+

89101instance Monoid Builder where

90102 mempty = empty

91103 {-# INLINE mempty #-}

104+#if MIN_VERSION_base(4,9,0)

105+ mappend = (<>) -- future-proof definition

106+#else

92107 mappend = append

108+#endif

93109 {-# INLINE mappend #-}

94− mconcat = foldr mappend mempty

110+ mconcat = foldr mappend Data.Monoid.mempty

95111 {-# INLINE mconcat #-}

96112

97113instance String.IsString Builder where

@@ -122,17 +138,7 @@ empty = Builder (\ k buf -> k buf)

122138-- * @'toLazyText' ('singleton' c) = 'L.singleton' c@

123139--

124140singleton :: Char -> Builder

125−singleton c = writeAtMost 2 $ \ marr o ->

126− if n < 0x10000

127− then A.unsafeWrite marr o (fromIntegral n) >> return 1

128− else do

129− A.unsafeWrite marr o lo

130− A.unsafeWrite marr (o+1) hi

131− return 2

132− where n = ord c

133− m = n - 0x10000

134− lo = fromIntegral $ (m `shiftR` 10) + 0xD800

135− hi = fromIntegral $ (m .&. 0x3FF) + 0xDC00

141+singleton c = writeAtMost 2 $ \ marr o -> unsafeWrite marr o c

136142{-# INLINE singleton #-}

137143

138144------------------------------------------------------------------------

@@ -236,6 +242,8 @@ flush = Builder $ \ k buf@(Buffer p o u l) ->

236242 !t = Text arr o u

237243 ts <- inlineInterleaveST (k b)

238244 return $! t : ts

245+{-# INLINE [1] flush #-}

246+-- defer inlining so that flush/flush rule may fire.

239247

240248------------------------------------------------------------------------

241249

Data/Text/Lazy/Builder/Functions.hs → Data/Text/Internal/Builder/Functions.hsrenamed+8−3

@@ -1,7 +1,7 @@

11{-# LANGUAGE MagicHash #-}

33-- |

4−-- Module : Data.Text.Lazy.Builder.Functions

4+-- Module : Data.Text.Internal.Builder.Functions

66--

77-- License : BSD-style

@@ -9,9 +9,13 @@

99-- Stability : experimental

1010-- Portability : GHC

1111--

12+-- /Warning/: this is an internal module, and does not have a stable

13+-- API or name. Functions in this module may not check or enforce

14+-- preconditions expected by public modules. Use at your own risk!

15+--

1216-- Useful functions and combinators.

1317

14−module Data.Text.Lazy.Builder.Functions

18+module Data.Text.Internal.Builder.Functions

1519 (

1620 (<>)

1721 , i2d

@@ -19,7 +23,8 @@ module Data.Text.Lazy.Builder.Functions

1923

2024import Data.Monoid (mappend)

2125import Data.Text.Lazy.Builder (Builder)

22−import GHC.Base

26+import GHC.Base (chr#,ord#,(+#),Int(I#),Char(C#))

27+import Prelude ()

2328

2429-- | Unsafe conversion for decimal digits.

2530{-# INLINE i2d #-}

Data/Text/Lazy/Builder/Int/Digits.hs → Data/Text/Internal/Builder/Int/Digits.hsrenamed+7−3

@@ -1,16 +1,20 @@

11{-# LANGUAGE OverloadedStrings #-}

3−-- Module: Data.Text.Lazy.Builder.Int.Digits

3+-- Module: Data.Text.Internal.Builder.Int.Digits

5−-- License: BSD3

5+-- License: BSD-style

66-- Maintainer: Bryan O'Sullivan <bos@serpentine.com>

77-- Stability: experimental

88-- Portability: portable

99--

10+-- /Warning/: this is an internal module, and does not have a stable

11+-- API or name. Functions in this module may not check or enforce

12+-- preconditions expected by public modules. Use at your own risk!

13+--

1014-- This module exists because the C preprocessor does things that we

1115-- shall not speak of when confronted with Haskell multiline strings.

1216

13−module Data.Text.Lazy.Builder.Int.Digits (digits) where

17+module Data.Text.Internal.Builder.Int.Digits (digits) where

1418

1519import Data.ByteString.Char8 (ByteString)

1620

Data/Text/Internal/Builder/RealFloat/Functions.hsnew file+57

@@ -0,0 +1,57 @@

1+{-# LANGUAGE CPP #-}

3+-- |

4+-- Module: Data.Text.Internal.Builder.RealFloat.Functions

5+-- Copyright: (c) The University of Glasgow 1994-2002

6+-- License: see libraries/base/LICENSE

7+--

8+-- /Warning/: this is an internal module, and does not have a stable

9+-- API or name. Functions in this module may not check or enforce

10+-- preconditions expected by public modules. Use at your own risk!

11+

12+module Data.Text.Internal.Builder.RealFloat.Functions

13+ (

14+ roundTo

15+ ) where

16+

17+roundTo :: Int -> [Int] -> (Int,[Int])

18+

19+#if MIN_VERSION_base(4,6,0)

20+

21+roundTo d is =

22+ case f d True is of

23+ x@(0,_) -> x

24+ (1,xs) -> (1, 1:xs)

25+ _ -> error "roundTo: bad Value"

26+ where

27+ b2 = base `quot` 2

28+

29+ f n _ [] = (0, replicate n 0)

30+ f 0 e (x:xs) | x == b2 && e && all (== 0) xs = (0, []) -- Round to even when at exactly half the base

31+ | otherwise = (if x >= b2 then 1 else 0, [])

32+ f n _ (i:xs)

33+ | i' == base = (1,0:ds)

34+ | otherwise = (0,i':ds)

35+ where

36+ (c,ds) = f (n-1) (even i) xs

37+ i' = c + i

38+ base = 10

39+

40+#else

41+

42+roundTo d is =

43+ case f d is of

44+ x@(0,_) -> x

45+ (1,xs) -> (1, 1:xs)

46+ _ -> error "roundTo: bad Value"

47+ where

48+ f n [] = (0, replicate n 0)

49+ f 0 (x:_) = (if x >= 5 then 1 else 0, [])

50+ f n (i:xs)

51+ | i' == 10 = (1,0:ds)

52+ | otherwise = (0,i':ds)

53+ where

54+ (c,ds) = f (n-1) xs

55+ i' = c + i

56+

57+#endif

Data/Text/Encoding/Fusion.hs → Data/Text/Internal/Encoding/Fusion.hsrenamed+17−14

@@ -1,21 +1,24 @@

11{-# LANGUAGE BangPatterns, CPP, Rank2Types #-}

33-- |

4−-- Module : Data.Text.Encoding.Fusion

4+-- Module : Data.Text.Internal.Encoding.Fusion

55-- Copyright : (c) Tom Harper 2008-2009,

66-- (c) Bryan O'Sullivan 2009,

77-- (c) Duncan Coutts 2009

88--

99-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

10+-- Maintainer : bos@serpentine.com

1211-- Stability : experimental

1312-- Portability : portable

1413--

14+-- /Warning/: this is an internal module, and does not have a stable

15+-- API or name. Functions in this module may not check or enforce

16+-- preconditions expected by public modules. Use at your own risk!

17+--

1518-- Fusible 'Stream'-oriented functions for converting between 'Text'

1619-- and several common encodings.

1720

18−module Data.Text.Encoding.Fusion

21+module Data.Text.Internal.Encoding.Fusion

1922 (

2023 -- * Streaming

2124 streamASCII

@@ -28,27 +31,27 @@ module Data.Text.Encoding.Fusion

2831 -- * Unstreaming

2932 , unstream

3033

31− , module Data.Text.Encoding.Fusion.Common

34+ , module Data.Text.Internal.Encoding.Fusion.Common

3235 ) where

3336

3437#if defined(ASSERTS)

3538import Control.Exception (assert)

3639#endif

3740import Data.ByteString.Internal (ByteString(..), mallocByteString, memcpy)

38−import Data.Text.Fusion (Step(..), Stream(..))

39−import Data.Text.Fusion.Size

41+import Data.Text.Internal.Fusion (Step(..), Stream(..))

42+import Data.Text.Internal.Fusion.Size

4043import Data.Text.Encoding.Error

41−import Data.Text.Encoding.Fusion.Common

42−import Data.Text.UnsafeChar (unsafeChr, unsafeChr8, unsafeChr32)

43−import Data.Text.UnsafeShift (shiftL, shiftR)

44+import Data.Text.Internal.Encoding.Fusion.Common

45+import Data.Text.Internal.Unsafe.Char (unsafeChr, unsafeChr8, unsafeChr32)

46+import Data.Text.Internal.Unsafe.Shift (shiftL, shiftR)

4447import Data.Word (Word8, Word16, Word32)

4548import Foreign.ForeignPtr (withForeignPtr, ForeignPtr)

4649import Foreign.Storable (pokeByteOff)

4750import qualified Data.ByteString as B

4851import qualified Data.ByteString.Unsafe as B

49−import qualified Data.Text.Encoding.Utf8 as U8

50−import qualified Data.Text.Encoding.Utf16 as U16

51−import qualified Data.Text.Encoding.Utf32 as U32

52+import qualified Data.Text.Internal.Encoding.Utf8 as U8

53+import qualified Data.Text.Internal.Encoding.Utf16 as U16

54+import qualified Data.Text.Internal.Encoding.Utf32 as U32

5255import Data.Text.Unsafe (unsafeDupablePerformIO)

5356

5457streamASCII :: ByteString -> Stream Char

@@ -201,5 +204,5 @@ decodeError func kind onErr mb i =

201204 case onErr desc mb of

202205 Nothing -> Skip i

203206 Just c -> Yield c i

204− where desc = "Data.Text.Encoding.Fusion." ++ func ++ ": Invalid " ++

207+ where desc = "Data.Text.Internal.Encoding.Fusion." ++ func ++ ": Invalid " ++

205208 kind ++ " stream"

Data/Text/Encoding/Fusion/Common.hs → Data/Text/Internal/Encoding/Fusion/Common.hsrenamed+10−8

@@ -1,22 +1,24 @@

11{-# LANGUAGE BangPatterns #-}

33-- |

4−-- Module : Data.Text.Encoding.Fusion.Common

4+-- Module : Data.Text.Internal.Encoding.Fusion.Common

55-- Copyright : (c) Tom Harper 2008-2009,

66-- (c) Bryan O'Sullivan 2009,

77-- (c) Duncan Coutts 2009,

88-- (c) Jasper Van der Jeugt 2011

99--

1010-- License : BSD-style

11−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

12−-- duncan@haskell.org

11+-- Maintainer : bos@serpentine.com

1312-- Stability : experimental

1413-- Portability : portable

1514--

15+-- /Warning/: this is an internal module, and does not have a stable

16+-- API or name. Use at your own risk!

17+--

1618-- Fusible 'Stream'-oriented functions for converting between 'Text'

1719-- and several common encodings.

1820

19−module Data.Text.Encoding.Fusion.Common

21+module Data.Text.Internal.Encoding.Fusion.Common

2022 (

2123 -- * Restreaming

2224 -- Restreaming is the act of converting from one 'Stream'

@@ -28,10 +30,10 @@ module Data.Text.Encoding.Fusion.Common

2830 ) where

2931

3032import Data.Bits ((.&.))

31−import Data.Text.Fusion (Step(..), Stream(..))

32−import Data.Text.Fusion.Internal (RS(..))

33−import Data.Text.UnsafeChar (ord)

34−import Data.Text.UnsafeShift (shiftR)

33+import Data.Text.Internal.Fusion (Step(..), Stream(..))

34+import Data.Text.Internal.Fusion.Types (RS(..))

35+import Data.Text.Internal.Unsafe.Char (ord)

36+import Data.Text.Internal.Unsafe.Shift (shiftR)

3537import Data.Word (Word8)

3638

3739restreamUtf16BE :: Stream Char -> Stream Word8

Data/Text/Encoding/Utf16.hs → Data/Text/Internal/Encoding/Utf16.hsrenamed+7−4

@@ -1,19 +1,22 @@

11{-# LANGUAGE MagicHash, BangPatterns #-}

33-- |

4−-- Module : Data.Text.Encoding.Utf16

4+-- Module : Data.Text.Internal.Encoding.Utf16

88--

99-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

10+-- Maintainer : bos@serpentine.com

1211-- Stability : experimental

1312-- Portability : GHC

1413--

14+-- /Warning/: this is an internal module, and does not have a stable

15+-- API or name. Functions in this module may not check or enforce

16+-- preconditions expected by public modules. Use at your own risk!

17+--

1518-- Basic UTF-16 validation and character manipulation.

16−module Data.Text.Encoding.Utf16

19+module Data.Text.Internal.Encoding.Utf16

1720 (

1821 chr2

1922 , validate1

Data/Text/Encoding/Utf32.hs → Data/Text/Internal/Encoding/Utf32.hsrenamed+7−4

@@ -1,17 +1,20 @@

11-- |

2−-- Module : Data.Text.Encoding.Utf32

2+-- Module : Data.Text.Internal.Encoding.Utf32

66--

77-- License : BSD-style

8−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

9−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

109-- Stability : experimental

1110-- Portability : portable

1211--

12+-- /Warning/: this is an internal module, and does not have a stable

13+-- API or name. Functions in this module may not check or enforce

14+-- preconditions expected by public modules. Use at your own risk!

15+--

1316-- Basic UTF-32 validation.

14−module Data.Text.Encoding.Utf32

17+module Data.Text.Internal.Encoding.Utf32

1518 (

1619 validate

1720 ) where

Data/Text/Encoding/Utf8.hs → Data/Text/Internal/Encoding/Utf8.hsrenamed+13−6

@@ -1,19 +1,22 @@

11{-# LANGUAGE CPP, MagicHash, BangPatterns #-}

33-- |

4−-- Module : Data.Text.Encoding.Utf8

4+-- Module : Data.Text.Internal.Encoding.Utf8

88--

99-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

10+-- Maintainer : bos@serpentine.com

1211-- Stability : experimental

1312-- Portability : GHC

1413--

14+-- /Warning/: this is an internal module, and does not have a stable

15+-- API or name. Functions in this module may not check or enforce

16+-- preconditions expected by public modules. Use at your own risk!

17+--

1518-- Basic UTF-8 validation and character manipulation.

16−module Data.Text.Encoding.Utf8

19+module Data.Text.Internal.Encoding.Utf8

1720 (

1821 -- Decomposition

1922 ord2

@@ -30,12 +33,16 @@ module Data.Text.Encoding.Utf8

3033 , validate4

3134 ) where

3235

36+#if defined(TEST_SUITE)

37+# undef ASSERTS

38+#endif

39+

3340#if defined(ASSERTS)

3441import Control.Exception (assert)

3542#endif

3643import Data.Bits ((.&.))

37−import Data.Text.UnsafeChar (ord)

38−import Data.Text.UnsafeShift (shiftR)

44+import Data.Text.Internal.Unsafe.Char (ord)

45+import Data.Text.Internal.Unsafe.Shift (shiftR)

3946import GHC.Exts

4047import GHC.Word (Word8(..))

4148

Data/Text/Util.hs → Data/Text/Internal/Functions.hsrenamed+6−2

@@ -1,7 +1,7 @@

11{-# LANGUAGE CPP, DeriveDataTypeable #-}

33-- |

4−-- Module : Data.Text.Util

4+-- Module : Data.Text.Internal.Functions

55-- Copyright : 2010 Bryan O'Sullivan

66--

77-- License : BSD-style

@@ -9,9 +9,13 @@

99-- Stability : experimental

1010-- Portability : GHC

1111--

12+-- /Warning/: this is an internal module, and does not have a stable

13+-- API or name. Functions in this module may not check or enforce

14+-- preconditions expected by public modules. Use at your own risk!

15+--

1216-- Useful functions.

1317

14−module Data.Text.Util

18+module Data.Text.Internal.Functions

1519 (

1620 intersperse

1721 ) where

Data/Text/Fusion.hs → Data/Text/Internal/Fusion.hsrenamed+52−39

@@ -1,20 +1,23 @@

11{-# LANGUAGE BangPatterns, MagicHash #-}

33-- |

4−-- Module : Data.Text.Fusion

4+-- Module : Data.Text.Internal.Fusion

55-- Copyright : (c) Tom Harper 2008-2009,

66-- (c) Bryan O'Sullivan 2009-2010,

77-- (c) Duncan Coutts 2009

88--

99-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

10+-- Maintainer : bos@serpentine.com

1211-- Stability : experimental

1312-- Portability : GHC

1413--

14+-- /Warning/: this is an internal module, and does not have a stable

15+-- API or name. Functions in this module may not check or enforce

16+-- preconditions expected by public modules. Use at your own risk!

17+--

1518-- Text manipulation functions represented as fusible operations over

1619-- streams.

17−module Data.Text.Fusion

20+module Data.Text.Internal.Fusion

1821 (

1922 -- * Types

2023 Stream(..)

@@ -51,21 +54,21 @@ import Prelude (Bool(..), Char, Maybe(..), Monad(..), Int,

5154 fromIntegral, otherwise)

5255import Data.Bits ((.&.))

5356import Data.Text.Internal (Text(..))

54−import Data.Text.Private (runText)

55−import Data.Text.UnsafeChar (ord, unsafeChr, unsafeWrite)

56−import Data.Text.UnsafeShift (shiftL, shiftR)

57+import Data.Text.Internal.Private (runText)

58+import Data.Text.Internal.Unsafe.Char (ord, unsafeChr, unsafeWrite)

59+import Data.Text.Internal.Unsafe.Shift (shiftL, shiftR)

5760import qualified Data.Text.Array as A

58−import qualified Data.Text.Fusion.Common as S

59−import Data.Text.Fusion.Internal

60−import Data.Text.Fusion.Size

61+import qualified Data.Text.Internal.Fusion.Common as S

62+import Data.Text.Internal.Fusion.Types

63+import Data.Text.Internal.Fusion.Size

6164import qualified Data.Text.Internal as I

62−import qualified Data.Text.Encoding.Utf16 as U16

65+import qualified Data.Text.Internal.Encoding.Utf16 as U16

6366

6467default(Int)

6568

6669-- | /O(n)/ Convert a 'Text' into a 'Stream Char'.

6770stream :: Text -> Stream Char

68−stream (Text arr off len) = Stream next off (maxSize len)

71+stream (Text arr off len) = Stream next off (betweenSize (len `shiftR` 1) len)

6972 where

7073 !end = off+len

7174 next !i

@@ -80,7 +83,7 @@ stream (Text arr off len) = Stream next off (maxSize len)

8083-- | /O(n)/ Convert a 'Text' into a 'Stream Char', but iterate

8184-- backwards.

8285reverseStream :: Text -> Stream Char

83−reverseStream (Text arr off len) = Stream next (off+len-1) (maxSize len)

86+reverseStream (Text arr off len) = Stream next (off+len-1) (betweenSize (len `shiftR` 1) len)

8487 where

8588 {-# INLINE next #-}

8689 next !i

@@ -95,25 +98,35 @@ reverseStream (Text arr off len) = Stream next (off+len-1) (maxSize len)

9598-- | /O(n)/ Convert a 'Stream Char' into a 'Text'.

9699unstream :: Stream Char -> Text

97100unstream (Stream next0 s0 len) = runText $ \done -> do

98− let mlen = upperBound 4 len

101+ -- Before encoding each char we perform a buffer realloc check assuming

102+ -- worst case encoding size of two 16-bit units for the char. Just add an

103+ -- extra space to the buffer so that we do not end up reallocating even when

104+ -- all the chars are encoded as single unit.

105+ let mlen = upperBound 4 len + 1

99106 arr0 <- A.new mlen

100− let outer arr top = loop

107+ let outer !arr !maxi = encode

101108 where

102− loop !s !i =

103− case next0 s of

104− Done -> done arr i

105− Skip s' -> loop s' i

106− Yield x s'

107− | j >= top -> {-# SCC "unstream/resize" #-} do

108− let top' = (top + 1) `shiftL` 1

109− arr' <- A.new top'

110− A.copyM arr' 0 arr 0 top

111− outer arr' top' s i

112− | otherwise -> do d <- unsafeWrite arr i x

113− loop s' (i+d)

114− where j | ord x < 0x10000 = i

115− | otherwise = i + 1

116− outer arr0 mlen s0 0

109+ -- keep the common case loop as small as possible

110+ encode !si !di =

111+ case next0 si of

112+ Done -> done arr di

113+ Skip si' -> encode si' di

114+ Yield c si'

115+ -- simply check for the worst case

116+ | maxi < di + 1 -> realloc si di

117+ | otherwise -> do

118+ n <- unsafeWrite arr di c

119+ encode si' (di + n)

120+

121+ -- keep uncommon case separate from the common case code

122+ {-# NOINLINE realloc #-}

123+ realloc !si !di = do

124+ let newlen = (maxi + 1) * 2

125+ arr' <- A.new newlen

126+ A.copyM arr' 0 arr 0 di

127+ outer arr' (newlen - 1) si di

128+

129+ outer arr0 (mlen - 1) s0 0

117130{-# INLINE [0] unstream #-}

118131{-# RULES "STREAM stream/unstream fusion" forall s. stream (unstream s) = s #-}

119132

@@ -129,7 +142,7 @@ length = S.lengthI

129142reverse :: Stream Char -> Text

130143reverse (Stream next s len0)

131144 | isEmpty len0 = I.empty

132− | otherwise = I.textP arr off' len'

145+ | otherwise = I.text arr off' len'

133146 where

134147 len0' = upperBound 4 (larger len0 4)

135148 (arr, (off', len')) = A.run2 (A.new len0' >>= loop s (len0'-1) len0')

@@ -163,15 +176,15 @@ reverse (Stream next s len0)

163176-- | /O(n)/ Perform the equivalent of 'scanr' over a list, only with

164177-- the input and result reversed.

165178reverseScanr :: (Char -> Char -> Char) -> Char -> Stream Char -> Stream Char

166−reverseScanr f z0 (Stream next0 s0 len) = Stream next (S1 :*: z0 :*: s0) (len+1) -- HINT maybe too low

179+reverseScanr f z0 (Stream next0 s0 len) = Stream next (Scan1 z0 s0) (len+1) -- HINT maybe too low

167180 where

168181 {-# INLINE next #-}

169− next (S1 :*: z :*: s) = Yield z (S2 :*: z :*: s)

170− next (S2 :*: z :*: s) = case next0 s of

171− Yield x s' -> let !x' = f x z

172− in Yield x' (S2 :*: x' :*: s')

173− Skip s' -> Skip (S2 :*: z :*: s')

174− Done -> Done

182+ next (Scan1 z s) = Yield z (Scan2 z s)

183+ next (Scan2 z s) = case next0 s of

184+ Yield x s' -> let !x' = f x z

185+ in Yield x' (Scan2 x' s')

186+ Skip s' -> Skip (Scan2 z s')

187+ Done -> Done

175188{-# INLINE reverseScanr #-}

176189

177190-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a stream from a seed

@@ -207,7 +220,7 @@ countChar = S.countCharI

207220-- function to each element of a 'Text', passing an accumulating

208221-- parameter from left to right, and returns a final 'Text'.

209222mapAccumL :: (a -> Char -> (a,Char)) -> a -> Stream Char -> (a, Text)

210−mapAccumL f z0 (Stream next0 s0 len) = (nz,I.textP na 0 nl)

223+mapAccumL f z0 (Stream next0 s0 len) = (nz, I.text na 0 nl)

211224 where

212225 (na,(nz,nl)) = A.run2 (A.new mlen >>= \arr -> outer arr mlen z0 s0 0)

213226 where mlen = upperBound 4 len

Data/Text/Fusion/CaseMapping.hs → Data/Text/Internal/Fusion/CaseMapping.hsrenamed+293−21

@@ -1,17 +1,17 @@

11{-# LANGUAGE Rank2Types #-}

22-- AUTOMATICALLY GENERATED - DO NOT EDIT

3−-- Generated by scripts/SpecialCasing.hs

4−-- CaseFolding-6.3.0.txt

5−-- Date: 2012-12-20, 22:14:35 GMT [MD]

6−-- SpecialCasing-6.3.0.txt

7−-- Date: 2013-05-08, 13:54:51 GMT [MD]

3+-- Generated by scripts/CaseMapping.hs

4+-- CaseFolding-8.0.0.txt

5+-- Date: 2015-01-13, 18:16:36 GMT [MD]

6+-- SpecialCasing-8.0.0.txt

7+-- Date: 2014-12-16, 23:08:04 GMT [MD]

9−module Data.Text.Fusion.CaseMapping where

9+module Data.Text.Internal.Fusion.CaseMapping where

1010import Data.Char

11−import Data.Text.Fusion.Internal

11+import Data.Text.Internal.Fusion.Types

1212

1313upperMapping :: forall s. Char -> s -> Step (CC s) Char

14−{-# INLINE upperMapping #-}

14+{-# NOINLINE upperMapping #-}

1515-- LATIN SMALL LETTER SHARP S

1616upperMapping '\x00df' s = Yield '\x0053' (CC s '\x0053' '\x0000')

1717-- LATIN SMALL LIGATURE FF

@@ -218,12 +218,12 @@ upperMapping '\x1fc7' s = Yield '\x0397' (CC s '\x0342' '\x0399')

218218upperMapping '\x1ff7' s = Yield '\x03a9' (CC s '\x0342' '\x0399')

219219upperMapping c s = Yield (toUpper c) (CC s '\0' '\0')

220220lowerMapping :: forall s. Char -> s -> Step (CC s) Char

221−{-# INLINE lowerMapping #-}

221+{-# NOINLINE lowerMapping #-}

222222-- LATIN CAPITAL LETTER I WITH DOT ABOVE

223223lowerMapping '\x0130' s = Yield '\x0069' (CC s '\x0307' '\x0000')

224224lowerMapping c s = Yield (toLower c) (CC s '\0' '\0')

225225titleMapping :: forall s. Char -> s -> Step (CC s) Char

226−{-# INLINE titleMapping #-}

226+{-# NOINLINE titleMapping #-}

227227-- LATIN SMALL LETTER SHARP S

228228titleMapping '\x00df' s = Yield '\x0053' (CC s '\x0073' '\x0000')

229229-- LATIN SMALL LIGATURE FF

@@ -322,7 +322,7 @@ titleMapping '\x1fc7' s = Yield '\x0397' (CC s '\x0342' '\x0345')

322322titleMapping '\x1ff7' s = Yield '\x03a9' (CC s '\x0342' '\x0345')

323323titleMapping c s = Yield (toTitle c) (CC s '\0' '\0')

324324foldMapping :: forall s. Char -> s -> Step (CC s) Char

325−{-# INLINE foldMapping #-}

325+{-# NOINLINE foldMapping #-}

326326-- MICRO SIGN

327327foldMapping '\x00b5' s = Yield '\x03bc' (CC s '\x0000' '\x0000')

328328-- LATIN SMALL LETTER SHARP S

@@ -359,10 +359,18 @@ foldMapping '\x03f1' s = Yield '\x03c1' (CC s '\x0000' '\x0000')

359359foldMapping '\x03f5' s = Yield '\x03b5' (CC s '\x0000' '\x0000')

360360-- ARMENIAN SMALL LIGATURE ECH YIWN

361361foldMapping '\x0587' s = Yield '\x0565' (CC s '\x0582' '\x0000')

362−-- GEORGIAN CAPITAL LETTER YN

363−foldMapping '\x10c7' s = Yield '\x2d27' (CC s '\x0000' '\x0000')

364−-- GEORGIAN CAPITAL LETTER AEN

365−foldMapping '\x10cd' s = Yield '\x2d2d' (CC s '\x0000' '\x0000')

362+-- CHEROKEE SMALL LETTER YE

363+foldMapping '\x13f8' s = Yield '\x13f0' (CC s '\x0000' '\x0000')

364+-- CHEROKEE SMALL LETTER YI

365+foldMapping '\x13f9' s = Yield '\x13f1' (CC s '\x0000' '\x0000')

366+-- CHEROKEE SMALL LETTER YO

367+foldMapping '\x13fa' s = Yield '\x13f2' (CC s '\x0000' '\x0000')

368+-- CHEROKEE SMALL LETTER YU

369+foldMapping '\x13fb' s = Yield '\x13f3' (CC s '\x0000' '\x0000')

370+-- CHEROKEE SMALL LETTER YV

371+foldMapping '\x13fc' s = Yield '\x13f4' (CC s '\x0000' '\x0000')

372+-- CHEROKEE SMALL LETTER MV

373+foldMapping '\x13fd' s = Yield '\x13f5' (CC s '\x0000' '\x0000')

366374-- LATIN SMALL LETTER H WITH LINE BELOW

367375foldMapping '\x1e96' s = Yield '\x0068' (CC s '\x0331' '\x0000')

368376-- LATIN SMALL LETTER T WITH DIAERESIS

@@ -537,12 +545,174 @@ foldMapping '\x1ff6' s = Yield '\x03c9' (CC s '\x0342' '\x0000')

537545foldMapping '\x1ff7' s = Yield '\x03c9' (CC s '\x0342' '\x03b9')

538546-- GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI

539547foldMapping '\x1ffc' s = Yield '\x03c9' (CC s '\x03b9' '\x0000')

540−-- COPTIC CAPITAL LETTER BOHAIRIC KHEI

541−foldMapping '\x2cf2' s = Yield '\x2cf3' (CC s '\x0000' '\x0000')

542−-- LATIN CAPITAL LETTER C WITH BAR

543−foldMapping '\xa792' s = Yield '\xa793' (CC s '\x0000' '\x0000')

544−-- LATIN CAPITAL LETTER H WITH HOOK

545−foldMapping '\xa7aa' s = Yield '\x0266' (CC s '\x0000' '\x0000')

548+-- LATIN CAPITAL LETTER J WITH CROSSED-TAIL

549+foldMapping '\xa7b2' s = Yield '\x029d' (CC s '\x0000' '\x0000')

550+-- LATIN CAPITAL LETTER CHI

551+foldMapping '\xa7b3' s = Yield '\xab53' (CC s '\x0000' '\x0000')

552+-- LATIN CAPITAL LETTER BETA

553+foldMapping '\xa7b4' s = Yield '\xa7b5' (CC s '\x0000' '\x0000')

554+-- LATIN CAPITAL LETTER OMEGA

555+foldMapping '\xa7b6' s = Yield '\xa7b7' (CC s '\x0000' '\x0000')

556+-- CHEROKEE SMALL LETTER A

557+foldMapping '\xab70' s = Yield '\x13a0' (CC s '\x0000' '\x0000')

558+-- CHEROKEE SMALL LETTER E

559+foldMapping '\xab71' s = Yield '\x13a1' (CC s '\x0000' '\x0000')

560+-- CHEROKEE SMALL LETTER I

561+foldMapping '\xab72' s = Yield '\x13a2' (CC s '\x0000' '\x0000')

562+-- CHEROKEE SMALL LETTER O

563+foldMapping '\xab73' s = Yield '\x13a3' (CC s '\x0000' '\x0000')

564+-- CHEROKEE SMALL LETTER U

565+foldMapping '\xab74' s = Yield '\x13a4' (CC s '\x0000' '\x0000')

566+-- CHEROKEE SMALL LETTER V

567+foldMapping '\xab75' s = Yield '\x13a5' (CC s '\x0000' '\x0000')

568+-- CHEROKEE SMALL LETTER GA

569+foldMapping '\xab76' s = Yield '\x13a6' (CC s '\x0000' '\x0000')

570+-- CHEROKEE SMALL LETTER KA

571+foldMapping '\xab77' s = Yield '\x13a7' (CC s '\x0000' '\x0000')

572+-- CHEROKEE SMALL LETTER GE

573+foldMapping '\xab78' s = Yield '\x13a8' (CC s '\x0000' '\x0000')

574+-- CHEROKEE SMALL LETTER GI

575+foldMapping '\xab79' s = Yield '\x13a9' (CC s '\x0000' '\x0000')

576+-- CHEROKEE SMALL LETTER GO

577+foldMapping '\xab7a' s = Yield '\x13aa' (CC s '\x0000' '\x0000')

578+-- CHEROKEE SMALL LETTER GU

579+foldMapping '\xab7b' s = Yield '\x13ab' (CC s '\x0000' '\x0000')

580+-- CHEROKEE SMALL LETTER GV

581+foldMapping '\xab7c' s = Yield '\x13ac' (CC s '\x0000' '\x0000')

582+-- CHEROKEE SMALL LETTER HA

583+foldMapping '\xab7d' s = Yield '\x13ad' (CC s '\x0000' '\x0000')

584+-- CHEROKEE SMALL LETTER HE

585+foldMapping '\xab7e' s = Yield '\x13ae' (CC s '\x0000' '\x0000')

586+-- CHEROKEE SMALL LETTER HI

587+foldMapping '\xab7f' s = Yield '\x13af' (CC s '\x0000' '\x0000')

588+-- CHEROKEE SMALL LETTER HO

589+foldMapping '\xab80' s = Yield '\x13b0' (CC s '\x0000' '\x0000')

590+-- CHEROKEE SMALL LETTER HU

591+foldMapping '\xab81' s = Yield '\x13b1' (CC s '\x0000' '\x0000')

592+-- CHEROKEE SMALL LETTER HV

593+foldMapping '\xab82' s = Yield '\x13b2' (CC s '\x0000' '\x0000')

594+-- CHEROKEE SMALL LETTER LA

595+foldMapping '\xab83' s = Yield '\x13b3' (CC s '\x0000' '\x0000')

596+-- CHEROKEE SMALL LETTER LE

597+foldMapping '\xab84' s = Yield '\x13b4' (CC s '\x0000' '\x0000')

598+-- CHEROKEE SMALL LETTER LI

599+foldMapping '\xab85' s = Yield '\x13b5' (CC s '\x0000' '\x0000')

600+-- CHEROKEE SMALL LETTER LO

601+foldMapping '\xab86' s = Yield '\x13b6' (CC s '\x0000' '\x0000')

602+-- CHEROKEE SMALL LETTER LU

603+foldMapping '\xab87' s = Yield '\x13b7' (CC s '\x0000' '\x0000')

604+-- CHEROKEE SMALL LETTER LV

605+foldMapping '\xab88' s = Yield '\x13b8' (CC s '\x0000' '\x0000')

606+-- CHEROKEE SMALL LETTER MA

607+foldMapping '\xab89' s = Yield '\x13b9' (CC s '\x0000' '\x0000')

608+-- CHEROKEE SMALL LETTER ME

609+foldMapping '\xab8a' s = Yield '\x13ba' (CC s '\x0000' '\x0000')

610+-- CHEROKEE SMALL LETTER MI

611+foldMapping '\xab8b' s = Yield '\x13bb' (CC s '\x0000' '\x0000')

612+-- CHEROKEE SMALL LETTER MO

613+foldMapping '\xab8c' s = Yield '\x13bc' (CC s '\x0000' '\x0000')

614+-- CHEROKEE SMALL LETTER MU

615+foldMapping '\xab8d' s = Yield '\x13bd' (CC s '\x0000' '\x0000')

616+-- CHEROKEE SMALL LETTER NA

617+foldMapping '\xab8e' s = Yield '\x13be' (CC s '\x0000' '\x0000')

618+-- CHEROKEE SMALL LETTER HNA

619+foldMapping '\xab8f' s = Yield '\x13bf' (CC s '\x0000' '\x0000')

620+-- CHEROKEE SMALL LETTER NAH

621+foldMapping '\xab90' s = Yield '\x13c0' (CC s '\x0000' '\x0000')

622+-- CHEROKEE SMALL LETTER NE

623+foldMapping '\xab91' s = Yield '\x13c1' (CC s '\x0000' '\x0000')

624+-- CHEROKEE SMALL LETTER NI

625+foldMapping '\xab92' s = Yield '\x13c2' (CC s '\x0000' '\x0000')

626+-- CHEROKEE SMALL LETTER NO

627+foldMapping '\xab93' s = Yield '\x13c3' (CC s '\x0000' '\x0000')

628+-- CHEROKEE SMALL LETTER NU

629+foldMapping '\xab94' s = Yield '\x13c4' (CC s '\x0000' '\x0000')

630+-- CHEROKEE SMALL LETTER NV

631+foldMapping '\xab95' s = Yield '\x13c5' (CC s '\x0000' '\x0000')

632+-- CHEROKEE SMALL LETTER QUA

633+foldMapping '\xab96' s = Yield '\x13c6' (CC s '\x0000' '\x0000')

634+-- CHEROKEE SMALL LETTER QUE

635+foldMapping '\xab97' s = Yield '\x13c7' (CC s '\x0000' '\x0000')

636+-- CHEROKEE SMALL LETTER QUI

637+foldMapping '\xab98' s = Yield '\x13c8' (CC s '\x0000' '\x0000')

638+-- CHEROKEE SMALL LETTER QUO

639+foldMapping '\xab99' s = Yield '\x13c9' (CC s '\x0000' '\x0000')

640+-- CHEROKEE SMALL LETTER QUU

641+foldMapping '\xab9a' s = Yield '\x13ca' (CC s '\x0000' '\x0000')

642+-- CHEROKEE SMALL LETTER QUV

643+foldMapping '\xab9b' s = Yield '\x13cb' (CC s '\x0000' '\x0000')

644+-- CHEROKEE SMALL LETTER SA

645+foldMapping '\xab9c' s = Yield '\x13cc' (CC s '\x0000' '\x0000')

646+-- CHEROKEE SMALL LETTER S

647+foldMapping '\xab9d' s = Yield '\x13cd' (CC s '\x0000' '\x0000')

648+-- CHEROKEE SMALL LETTER SE

649+foldMapping '\xab9e' s = Yield '\x13ce' (CC s '\x0000' '\x0000')

650+-- CHEROKEE SMALL LETTER SI

651+foldMapping '\xab9f' s = Yield '\x13cf' (CC s '\x0000' '\x0000')

652+-- CHEROKEE SMALL LETTER SO

653+foldMapping '\xaba0' s = Yield '\x13d0' (CC s '\x0000' '\x0000')

654+-- CHEROKEE SMALL LETTER SU

655+foldMapping '\xaba1' s = Yield '\x13d1' (CC s '\x0000' '\x0000')

656+-- CHEROKEE SMALL LETTER SV

657+foldMapping '\xaba2' s = Yield '\x13d2' (CC s '\x0000' '\x0000')

658+-- CHEROKEE SMALL LETTER DA

659+foldMapping '\xaba3' s = Yield '\x13d3' (CC s '\x0000' '\x0000')

660+-- CHEROKEE SMALL LETTER TA

661+foldMapping '\xaba4' s = Yield '\x13d4' (CC s '\x0000' '\x0000')

662+-- CHEROKEE SMALL LETTER DE

663+foldMapping '\xaba5' s = Yield '\x13d5' (CC s '\x0000' '\x0000')

664+-- CHEROKEE SMALL LETTER TE

665+foldMapping '\xaba6' s = Yield '\x13d6' (CC s '\x0000' '\x0000')

666+-- CHEROKEE SMALL LETTER DI

667+foldMapping '\xaba7' s = Yield '\x13d7' (CC s '\x0000' '\x0000')

668+-- CHEROKEE SMALL LETTER TI

669+foldMapping '\xaba8' s = Yield '\x13d8' (CC s '\x0000' '\x0000')

670+-- CHEROKEE SMALL LETTER DO

671+foldMapping '\xaba9' s = Yield '\x13d9' (CC s '\x0000' '\x0000')

672+-- CHEROKEE SMALL LETTER DU

673+foldMapping '\xabaa' s = Yield '\x13da' (CC s '\x0000' '\x0000')

674+-- CHEROKEE SMALL LETTER DV

675+foldMapping '\xabab' s = Yield '\x13db' (CC s '\x0000' '\x0000')

676+-- CHEROKEE SMALL LETTER DLA

677+foldMapping '\xabac' s = Yield '\x13dc' (CC s '\x0000' '\x0000')

678+-- CHEROKEE SMALL LETTER TLA

679+foldMapping '\xabad' s = Yield '\x13dd' (CC s '\x0000' '\x0000')

680+-- CHEROKEE SMALL LETTER TLE

681+foldMapping '\xabae' s = Yield '\x13de' (CC s '\x0000' '\x0000')

682+-- CHEROKEE SMALL LETTER TLI

683+foldMapping '\xabaf' s = Yield '\x13df' (CC s '\x0000' '\x0000')

684+-- CHEROKEE SMALL LETTER TLO

685+foldMapping '\xabb0' s = Yield '\x13e0' (CC s '\x0000' '\x0000')

686+-- CHEROKEE SMALL LETTER TLU

687+foldMapping '\xabb1' s = Yield '\x13e1' (CC s '\x0000' '\x0000')

688+-- CHEROKEE SMALL LETTER TLV

689+foldMapping '\xabb2' s = Yield '\x13e2' (CC s '\x0000' '\x0000')

690+-- CHEROKEE SMALL LETTER TSA

691+foldMapping '\xabb3' s = Yield '\x13e3' (CC s '\x0000' '\x0000')

692+-- CHEROKEE SMALL LETTER TSE

693+foldMapping '\xabb4' s = Yield '\x13e4' (CC s '\x0000' '\x0000')

694+-- CHEROKEE SMALL LETTER TSI

695+foldMapping '\xabb5' s = Yield '\x13e5' (CC s '\x0000' '\x0000')

696+-- CHEROKEE SMALL LETTER TSO

697+foldMapping '\xabb6' s = Yield '\x13e6' (CC s '\x0000' '\x0000')

698+-- CHEROKEE SMALL LETTER TSU

699+foldMapping '\xabb7' s = Yield '\x13e7' (CC s '\x0000' '\x0000')

700+-- CHEROKEE SMALL LETTER TSV

701+foldMapping '\xabb8' s = Yield '\x13e8' (CC s '\x0000' '\x0000')

702+-- CHEROKEE SMALL LETTER WA

703+foldMapping '\xabb9' s = Yield '\x13e9' (CC s '\x0000' '\x0000')

704+-- CHEROKEE SMALL LETTER WE

705+foldMapping '\xabba' s = Yield '\x13ea' (CC s '\x0000' '\x0000')

706+-- CHEROKEE SMALL LETTER WI

707+foldMapping '\xabbb' s = Yield '\x13eb' (CC s '\x0000' '\x0000')

708+-- CHEROKEE SMALL LETTER WO

709+foldMapping '\xabbc' s = Yield '\x13ec' (CC s '\x0000' '\x0000')

710+-- CHEROKEE SMALL LETTER WU

711+foldMapping '\xabbd' s = Yield '\x13ed' (CC s '\x0000' '\x0000')

712+-- CHEROKEE SMALL LETTER WV

713+foldMapping '\xabbe' s = Yield '\x13ee' (CC s '\x0000' '\x0000')

714+-- CHEROKEE SMALL LETTER YA

715+foldMapping '\xabbf' s = Yield '\x13ef' (CC s '\x0000' '\x0000')

546716-- LATIN SMALL LIGATURE FF

547717foldMapping '\xfb00' s = Yield '\x0066' (CC s '\x0066' '\x0000')

548718-- LATIN SMALL LIGATURE FI

@@ -567,4 +737,106 @@ foldMapping '\xfb15' s = Yield '\x0574' (CC s '\x056b' '\x0000')

567737foldMapping '\xfb16' s = Yield '\x057e' (CC s '\x0576' '\x0000')

568738-- ARMENIAN SMALL LIGATURE MEN XEH

569739foldMapping '\xfb17' s = Yield '\x0574' (CC s '\x056d' '\x0000')

740+-- OLD HUNGARIAN CAPITAL LETTER A

741+foldMapping '\x10c80' s = Yield '\x10cc0' (CC s '\x0000' '\x0000')

742+-- OLD HUNGARIAN CAPITAL LETTER AA

743+foldMapping '\x10c81' s = Yield '\x10cc1' (CC s '\x0000' '\x0000')

744+-- OLD HUNGARIAN CAPITAL LETTER EB

745+foldMapping '\x10c82' s = Yield '\x10cc2' (CC s '\x0000' '\x0000')

746+-- OLD HUNGARIAN CAPITAL LETTER AMB

747+foldMapping '\x10c83' s = Yield '\x10cc3' (CC s '\x0000' '\x0000')

748+-- OLD HUNGARIAN CAPITAL LETTER EC

749+foldMapping '\x10c84' s = Yield '\x10cc4' (CC s '\x0000' '\x0000')

750+-- OLD HUNGARIAN CAPITAL LETTER ENC

751+foldMapping '\x10c85' s = Yield '\x10cc5' (CC s '\x0000' '\x0000')

752+-- OLD HUNGARIAN CAPITAL LETTER ECS

753+foldMapping '\x10c86' s = Yield '\x10cc6' (CC s '\x0000' '\x0000')

754+-- OLD HUNGARIAN CAPITAL LETTER ED

755+foldMapping '\x10c87' s = Yield '\x10cc7' (CC s '\x0000' '\x0000')

756+-- OLD HUNGARIAN CAPITAL LETTER AND

757+foldMapping '\x10c88' s = Yield '\x10cc8' (CC s '\x0000' '\x0000')

758+-- OLD HUNGARIAN CAPITAL LETTER E

759+foldMapping '\x10c89' s = Yield '\x10cc9' (CC s '\x0000' '\x0000')

760+-- OLD HUNGARIAN CAPITAL LETTER CLOSE E

761+foldMapping '\x10c8a' s = Yield '\x10cca' (CC s '\x0000' '\x0000')

762+-- OLD HUNGARIAN CAPITAL LETTER EE

763+foldMapping '\x10c8b' s = Yield '\x10ccb' (CC s '\x0000' '\x0000')

764+-- OLD HUNGARIAN CAPITAL LETTER EF

765+foldMapping '\x10c8c' s = Yield '\x10ccc' (CC s '\x0000' '\x0000')

766+-- OLD HUNGARIAN CAPITAL LETTER EG

767+foldMapping '\x10c8d' s = Yield '\x10ccd' (CC s '\x0000' '\x0000')

768+-- OLD HUNGARIAN CAPITAL LETTER EGY

769+foldMapping '\x10c8e' s = Yield '\x10cce' (CC s '\x0000' '\x0000')

770+-- OLD HUNGARIAN CAPITAL LETTER EH

771+foldMapping '\x10c8f' s = Yield '\x10ccf' (CC s '\x0000' '\x0000')

772+-- OLD HUNGARIAN CAPITAL LETTER I

773+foldMapping '\x10c90' s = Yield '\x10cd0' (CC s '\x0000' '\x0000')

774+-- OLD HUNGARIAN CAPITAL LETTER II

775+foldMapping '\x10c91' s = Yield '\x10cd1' (CC s '\x0000' '\x0000')

776+-- OLD HUNGARIAN CAPITAL LETTER EJ

777+foldMapping '\x10c92' s = Yield '\x10cd2' (CC s '\x0000' '\x0000')

778+-- OLD HUNGARIAN CAPITAL LETTER EK

779+foldMapping '\x10c93' s = Yield '\x10cd3' (CC s '\x0000' '\x0000')

780+-- OLD HUNGARIAN CAPITAL LETTER AK

781+foldMapping '\x10c94' s = Yield '\x10cd4' (CC s '\x0000' '\x0000')

782+-- OLD HUNGARIAN CAPITAL LETTER UNK

783+foldMapping '\x10c95' s = Yield '\x10cd5' (CC s '\x0000' '\x0000')

784+-- OLD HUNGARIAN CAPITAL LETTER EL

785+foldMapping '\x10c96' s = Yield '\x10cd6' (CC s '\x0000' '\x0000')

786+-- OLD HUNGARIAN CAPITAL LETTER ELY

787+foldMapping '\x10c97' s = Yield '\x10cd7' (CC s '\x0000' '\x0000')

788+-- OLD HUNGARIAN CAPITAL LETTER EM

789+foldMapping '\x10c98' s = Yield '\x10cd8' (CC s '\x0000' '\x0000')

790+-- OLD HUNGARIAN CAPITAL LETTER EN

791+foldMapping '\x10c99' s = Yield '\x10cd9' (CC s '\x0000' '\x0000')

792+-- OLD HUNGARIAN CAPITAL LETTER ENY

793+foldMapping '\x10c9a' s = Yield '\x10cda' (CC s '\x0000' '\x0000')

794+-- OLD HUNGARIAN CAPITAL LETTER O

795+foldMapping '\x10c9b' s = Yield '\x10cdb' (CC s '\x0000' '\x0000')

796+-- OLD HUNGARIAN CAPITAL LETTER OO

797+foldMapping '\x10c9c' s = Yield '\x10cdc' (CC s '\x0000' '\x0000')

798+-- OLD HUNGARIAN CAPITAL LETTER NIKOLSBURG OE

799+foldMapping '\x10c9d' s = Yield '\x10cdd' (CC s '\x0000' '\x0000')

800+-- OLD HUNGARIAN CAPITAL LETTER RUDIMENTA OE

801+foldMapping '\x10c9e' s = Yield '\x10cde' (CC s '\x0000' '\x0000')

802+-- OLD HUNGARIAN CAPITAL LETTER OEE

803+foldMapping '\x10c9f' s = Yield '\x10cdf' (CC s '\x0000' '\x0000')

804+-- OLD HUNGARIAN CAPITAL LETTER EP

805+foldMapping '\x10ca0' s = Yield '\x10ce0' (CC s '\x0000' '\x0000')

806+-- OLD HUNGARIAN CAPITAL LETTER EMP

807+foldMapping '\x10ca1' s = Yield '\x10ce1' (CC s '\x0000' '\x0000')

808+-- OLD HUNGARIAN CAPITAL LETTER ER

809+foldMapping '\x10ca2' s = Yield '\x10ce2' (CC s '\x0000' '\x0000')

810+-- OLD HUNGARIAN CAPITAL LETTER SHORT ER

811+foldMapping '\x10ca3' s = Yield '\x10ce3' (CC s '\x0000' '\x0000')

812+-- OLD HUNGARIAN CAPITAL LETTER ES

813+foldMapping '\x10ca4' s = Yield '\x10ce4' (CC s '\x0000' '\x0000')

814+-- OLD HUNGARIAN CAPITAL LETTER ESZ

815+foldMapping '\x10ca5' s = Yield '\x10ce5' (CC s '\x0000' '\x0000')

816+-- OLD HUNGARIAN CAPITAL LETTER ET

817+foldMapping '\x10ca6' s = Yield '\x10ce6' (CC s '\x0000' '\x0000')

818+-- OLD HUNGARIAN CAPITAL LETTER ENT

819+foldMapping '\x10ca7' s = Yield '\x10ce7' (CC s '\x0000' '\x0000')

820+-- OLD HUNGARIAN CAPITAL LETTER ETY

821+foldMapping '\x10ca8' s = Yield '\x10ce8' (CC s '\x0000' '\x0000')

822+-- OLD HUNGARIAN CAPITAL LETTER ECH

823+foldMapping '\x10ca9' s = Yield '\x10ce9' (CC s '\x0000' '\x0000')

824+-- OLD HUNGARIAN CAPITAL LETTER U

825+foldMapping '\x10caa' s = Yield '\x10cea' (CC s '\x0000' '\x0000')

826+-- OLD HUNGARIAN CAPITAL LETTER UU

827+foldMapping '\x10cab' s = Yield '\x10ceb' (CC s '\x0000' '\x0000')

828+-- OLD HUNGARIAN CAPITAL LETTER NIKOLSBURG UE

829+foldMapping '\x10cac' s = Yield '\x10cec' (CC s '\x0000' '\x0000')

830+-- OLD HUNGARIAN CAPITAL LETTER RUDIMENTA UE

831+foldMapping '\x10cad' s = Yield '\x10ced' (CC s '\x0000' '\x0000')

832+-- OLD HUNGARIAN CAPITAL LETTER EV

833+foldMapping '\x10cae' s = Yield '\x10cee' (CC s '\x0000' '\x0000')

834+-- OLD HUNGARIAN CAPITAL LETTER EZ

835+foldMapping '\x10caf' s = Yield '\x10cef' (CC s '\x0000' '\x0000')

836+-- OLD HUNGARIAN CAPITAL LETTER EZS

837+foldMapping '\x10cb0' s = Yield '\x10cf0' (CC s '\x0000' '\x0000')

838+-- OLD HUNGARIAN CAPITAL LETTER ENT-SHAPED SIGN

839+foldMapping '\x10cb1' s = Yield '\x10cf1' (CC s '\x0000' '\x0000')

840+-- OLD HUNGARIAN CAPITAL LETTER US

841+foldMapping '\x10cb2' s = Yield '\x10cf2' (CC s '\x0000' '\x0000')

570842foldMapping c s = Yield (toLower c) (CC s '\0' '\0')

Data/Text/Fusion/Common.hs → Data/Text/Internal/Fusion/Common.hsrenamed+75−84

@@ -1,17 +1,20 @@

11{-# LANGUAGE BangPatterns, MagicHash, Rank2Types #-}

22-- |

3−-- Module : Data.Text.Fusion.Common

3+-- Module : Data.Text.Internal.Fusion.Common

44-- Copyright : (c) Bryan O'Sullivan 2009, 2012

55--

66-- License : BSD-style

7−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

8−-- duncan@haskell.org

7+-- Maintainer : bos@serpentine.com

98-- Stability : experimental

109-- Portability : GHC

1110--

11+-- /Warning/: this is an internal module, and does not have a stable

12+-- API or name. Functions in this module may not check or enforce

13+-- preconditions expected by public modules. Use at your own risk!

14+--

1215-- Common stream fusion functionality for text.

1316

14−module Data.Text.Fusion.Common

17+module Data.Text.Internal.Fusion.Common

1518 (

1619 -- * Creation and elimination

1720 singleton

@@ -68,9 +71,6 @@ module Data.Text.Fusion.Common

6871 -- ** Scans

6972 , scanl

7073

71− -- ** Accumulating maps

72− -- , mapAccumL

73−

7474 -- ** Generation and unfolding

7575 , replicateCharI

7676 , replicateI

@@ -103,16 +103,16 @@ module Data.Text.Fusion.Common

103103

104104import Prelude (Bool(..), Char, Eq(..), Int, Integral, Maybe(..),

105105 Ord(..), Ordering(..), String, (.), ($), (+), (-), (*), (++),

106− (&&), fromIntegral, not, otherwise)

106+ (&&), fromIntegral, otherwise)

107107import qualified Data.List as L

108108import qualified Prelude as P

109109import Data.Bits (shiftL)

110110import Data.Char (isLetter)

111111import Data.Int (Int64)

112−import Data.Text.Fusion.Internal

113−import Data.Text.Fusion.CaseMapping (foldMapping, lowerMapping, titleMapping,

112+import Data.Text.Internal.Fusion.Types

113+import Data.Text.Internal.Fusion.CaseMapping (foldMapping, lowerMapping, titleMapping,

114114 upperMapping)

115−import Data.Text.Fusion.Size

115+import Data.Text.Internal.Fusion.Size

116116import GHC.Prim (Addr#, chr#, indexCharOffAddr#, ord#)

117117import GHC.Types (Char(..), Int(..))

118118

@@ -120,7 +120,7 @@ singleton :: Char -> Stream Char

120120singleton c = Stream next False 1

121121 where next False = Yield c True

122122 next True = Done

123−{-# INLINE singleton #-}

123+{-# INLINE [0] singleton #-}

124124

125125streamList :: [a] -> Stream a

126126{-# INLINE [0] streamList #-}

@@ -184,6 +184,9 @@ cons !w (Stream next0 s0 len) = Stream next (C1 s0) (len+1)

184184 Yield x s' -> Yield x (C0 s')

185185{-# INLINE [0] cons #-}

186186

187+data Snoc a = N

188+ | J !a

189+

187190-- | /O(n)/ Adds a character to the end of a stream.

188191snoc :: Stream Char -> Char -> Stream Char

189192snoc (Stream next0 xs0 len) w = Stream next (J xs0) (len+1)

@@ -312,12 +315,13 @@ lengthI (Stream next s0 _len) = loop_length 0 s0

312315--

313316-- This function gives the same answer as comparing against the result

314317-- of 'lengthI', but can short circuit if the count of characters is

315−-- greater than the number, and hence be more efficient.

318+-- greater than the number or if the stream can't possibly be as long

319+-- as the number supplied, and hence be more efficient.

316320compareLengthI :: Integral a => Stream Char -> a -> Ordering

317321compareLengthI (Stream next s0 len) n =

318− case exactly len of

322+ case compareSize len (fromIntegral n) of

323+ Just o -> o

319324 Nothing -> loop_cmp 0 s0

320− Just i -> compare (fromIntegral i) n

321325 where

322326 loop_cmp !z s = case next s of

323327 Done -> compare z n

@@ -469,17 +473,20 @@ toTitle (Stream next0 s0 len) = Stream next (CC (False :*: s0) '\0' '\0') len

469473 next (CC s a b) = Yield a (CC s b '\0')

470474{-# INLINE [0] toTitle #-}

471475

476+data Justify i s = Just1 !i !s

477+ | Just2 !i !s

478+

472479justifyLeftI :: Integral a => a -> Char -> Stream Char -> Stream Char

473480justifyLeftI k c (Stream next0 s0 len) =

474− Stream next (s0 :*: S1 :*: 0) (larger (fromIntegral k) len)

481+ Stream next (Just1 0 s0) (larger (fromIntegral k) len)

475482 where

476− next (s :*: S1 :*: n) =

483+ next (Just1 n s) =

477484 case next0 s of

478− Done -> next (s :*: S2 :*: n)

479− Skip s' -> Skip (s' :*: S1 :*: n)

480− Yield x s' -> Yield x (s' :*: S1 :*: n+1)

481− next (s :*: S2 :*: n)

482− | n < k = Yield c (s :*: S2 :*: n+1)

485+ Done -> next (Just2 n s)

486+ Skip s' -> Skip (Just1 n s')

487+ Yield x s' -> Yield x (Just1 (n+1) s')

488+ next (Just2 n s)

489+ | n < k = Yield c (Just2 (n+1) s)

483490 | otherwise = Done

484491 {-# INLINE next #-}

485492{-# INLINE [0] justifyLeftI #-}

@@ -586,7 +593,7 @@ concatMap f = foldr (append . f) empty

586593{-# INLINE [0] concatMap #-}

587594

588595-- | /O(n)/ any @p @xs determines if any character in the stream

589−-- @xs@ satisifes the predicate @p@.

596+-- @xs@ satisfies the predicate @p@.

590597any :: (Char -> Bool) -> Stream Char -> Bool

591598any p (Stream next0 s0 _len) = loop_any s0

592599 where

@@ -598,7 +605,7 @@ any p (Stream next0 s0 _len) = loop_any s0

598605{-# INLINE [0] any #-}

599606

600607-- | /O(n)/ all @p @xs determines if all characters in the 'Text'

601−-- @xs@ satisify the predicate @p@.

608+-- @xs@ satisfy the predicate @p@.

602609all :: (Char -> Bool) -> Stream Char -> Bool

603610all p (Stream next0 s0 _len) = loop_all s0

604611 where

@@ -647,50 +654,27 @@ minimum (Stream next0 s0 _len) = loop0_minimum s0

647654-- * Building streams

648655

649656scanl :: (Char -> Char -> Char) -> Char -> Stream Char -> Stream Char

650−scanl f z0 (Stream next0 s0 len) = Stream next (S1 :*: z0 :*: s0) (len+1) -- HINT maybe too low

657+scanl f z0 (Stream next0 s0 len) = Stream next (Scan1 z0 s0) (len+1) -- HINT maybe too low

651658 where

652659 {-# INLINE next #-}

653− next (S1 :*: z :*: s) = Yield z (S2 :*: z :*: s)

654− next (S2 :*: z :*: s) = case next0 s of

655− Yield x s' -> let !x' = f z x

656− in Yield x' (S2 :*: x' :*: s')

657− Skip s' -> Skip (S2 :*: z :*: s')

658− Done -> Done

660+ next (Scan1 z s) = Yield z (Scan2 z s)

661+ next (Scan2 z s) = case next0 s of

662+ Yield x s' -> let !x' = f z x

663+ in Yield x' (Scan2 x' s')

664+ Skip s' -> Skip (Scan2 z s')

665+ Done -> Done

659666{-# INLINE [0] scanl #-}

660667

661−-- -----------------------------------------------------------------------------

662−-- ** Accumulating maps

663−

664−{-

665−-- | /O(n)/ Like a combination of 'map' and 'foldl'. Applies a

666−-- function to each element of a stream, passing an accumulating

667−-- parameter from left to right, and returns a final stream.

668−--

669−-- /Note/: Unlike the version over lists, this function does not

670−-- return a final value for the accumulator, because the nature of

671−-- streams precludes it.

672−mapAccumL :: (a -> b -> (a,b)) -> a -> Stream b -> Stream b

673−mapAccumL f z0 (Stream next0 s0 len) = Stream next (s0 :*: z0) len -- HINT depends on f

674− where

675− {-# INLINE next #-}

676− next (s :*: z) = case next0 s of

677− Yield x s' -> let (z',y) = f z x

678− in Yield y (s' :*: z')

679− Skip s' -> Skip (s' :*: z)

680− Done -> Done

681−{-# INLINE [0] mapAccumL #-}

682−-}

683−

684668-- -----------------------------------------------------------------------------

685669-- ** Generating and unfolding streams

686670

687671replicateCharI :: Integral a => a -> Char -> Stream Char

688−replicateCharI n c

672+replicateCharI !n !c

689673 | n < 0 = empty

690674 | otherwise = Stream next 0 (fromIntegral n) -- HINT maybe too low

691675 where

692− next i | i >= n = Done

693− | otherwise = Yield c (i + 1)

676+ next !i | i >= n = Done

677+ | otherwise = Yield c (i + 1)

694678{-# INLINE [0] replicateCharI #-}

695679

696680data RI s = RI !s {-# UNPACK #-} !Int64

@@ -755,24 +739,27 @@ take n0 (Stream next0 s0 len) =

755739 Yield x s' -> Yield x ((n-1) :*: s')

756740{-# INLINE [0] take #-}

757741

742+data Drop a s = NS !s

743+ | JS !a !s

744+

758745-- | /O(n)/ drop n, applied to a stream, returns the suffix of the

759746-- stream after the first @n@ characters, or the empty stream if @n@

760747-- is greater than the length of the stream.

761748drop :: Integral a => a -> Stream Char -> Stream Char

762749drop n0 (Stream next0 s0 len) =

763− Stream next (J n0 :*: s0) (len - fromIntegral (max 0 n0))

750+ Stream next (JS n0 s0) (len - fromIntegral (max 0 n0))

764751 where

765752 {-# INLINE next #-}

766− next (J n :*: s)

767− | n <= 0 = Skip (N :*: s)

753+ next (JS n s)

754+ | n <= 0 = Skip (NS s)

768755 | otherwise = case next0 s of

769756 Done -> Done

770− Skip s' -> Skip (J n :*: s')

771− Yield _ s' -> Skip (J (n-1) :*: s')

772− next (N :*: s) = case next0 s of

757+ Skip s' -> Skip (JS n s')

758+ Yield _ s' -> Skip (JS (n-1) s')

759+ next (NS s) = case next0 s of

773760 Done -> Done

774− Skip s' -> Skip (N :*: s')

775− Yield x s' -> Yield x (N :*: s')

761+ Skip s' -> Skip (NS s')

762+ Yield x s' -> Yield x (NS s')

776763{-# INLINE [0] drop #-}

777764

778765-- | takeWhile, applied to a predicate @p@ and a stream, returns the

@@ -790,18 +777,18 @@ takeWhile p (Stream next0 s0 len) = Stream next s0 len -- HINT maybe too high

790777

791778-- | dropWhile @p @xs returns the suffix remaining after takeWhile @p @xs.

792779dropWhile :: (Char -> Bool) -> Stream Char -> Stream Char

793−dropWhile p (Stream next0 s0 len) = Stream next (S1 :*: s0) len -- HINT maybe too high

780+dropWhile p (Stream next0 s0 len) = Stream next (L s0) len -- HINT maybe too high

794781 where

795782 {-# INLINE next #-}

796− next (S1 :*: s) = case next0 s of

783+ next (L s) = case next0 s of

797784 Done -> Done

798− Skip s' -> Skip (S1 :*: s')

799− Yield x s' | p x -> Skip (S1 :*: s')

800− | otherwise -> Yield x (S2 :*: s')

801− next (S2 :*: s) = case next0 s of

785+ Skip s' -> Skip (L s')

786+ Yield x s' | p x -> Skip (L s')

787+ | otherwise -> Yield x (R s')

788+ next (R s) = case next0 s of

802789 Done -> Done

803− Skip s' -> Skip (S2 :*: s')

804− Yield x s' -> Yield x (S2 :*: s')

790+ Skip s' -> Skip (R s')

791+ Yield x s' -> Yield x (R s')

805792{-# INLINE [0] dropWhile #-}

806793

807794-- | /O(n)/ The 'isPrefixOf' function takes two 'Stream's and returns

@@ -908,21 +895,25 @@ findIndicesI p (Stream next s0 _len) = loop_findIndex 0 s0

908895-------------------------------------------------------------------------------

909896-- * Zipping

910897

898+-- | Strict triple.

899+data Zip a b m = Z1 !a !b

900+ | Z2 !a !b !m

901+

911902-- | zipWith generalises 'zip' by zipping with the function given as

912903-- the first argument, instead of a tupling function.

913904zipWith :: (a -> a -> b) -> Stream a -> Stream a -> Stream b

914905zipWith f (Stream next0 sa0 len1) (Stream next1 sb0 len2) =

915− Stream next (sa0 :*: sb0 :*: N) (smaller len1 len2)

906+ Stream next (Z1 sa0 sb0) (smaller len1 len2)

916907 where

917− next (sa :*: sb :*: N) = case next0 sa of

918− Done -> Done

919− Skip sa' -> Skip (sa' :*: sb :*: N)

920− Yield a sa' -> Skip (sa' :*: sb :*: J a)

921−

922− next (sa' :*: sb :*: J a) = case next1 sb of

923− Done -> Done

924− Skip sb' -> Skip (sa' :*: sb' :*: J a)

925− Yield b sb' -> Yield (f a b) (sa' :*: sb' :*: N)

908+ next (Z1 sa sb) = case next0 sa of

909+ Done -> Done

910+ Skip sa' -> Skip (Z1 sa' sb)

911+ Yield a sa' -> Skip (Z2 sa' sb a)

912+

913+ next (Z2 sa' sb a) = case next1 sb of

914+ Done -> Done

915+ Skip sb' -> Skip (Z2 sa' sb' a)

916+ Yield b sb' -> Yield (f a b) (Z1 sa' sb')

926917{-# INLINE [0] zipWith #-}

927918

928919-- | /O(n)/ The 'countCharI' function returns the number of times the

@@ -938,7 +929,7 @@ countCharI a (Stream next s0 _len) = loop 0 s0

938929{-# INLINE [0] countCharI #-}

939930

940931streamError :: String -> String -> a

941−streamError func msg = P.error $ "Data.Text.Fusion.Common." ++ func ++ ": " ++ msg

932+streamError func msg = P.error $ "Data.Text.Internal.Fusion.Common." ++ func ++ ": " ++ msg

942933

943934emptyError :: String -> a

944935emptyError func = internalError func "Empty input"

Data/Text/Internal/Fusion/Size.hsnew file+157

@@ -0,0 +1,157 @@

1+{-# LANGUAGE CPP #-}

2+{-# OPTIONS_GHC -fno-warn-missing-methods #-}

3+-- |

4+-- Module : Data.Text.Internal.Fusion.Internal

5+-- Copyright : (c) Roman Leshchinskiy 2008,

6+-- (c) Bryan O'Sullivan 2009

7+--

8+-- License : BSD-style

9+-- Maintainer : bos@serpentine.com

10+-- Stability : experimental

11+-- Portability : portable

12+--

13+-- /Warning/: this is an internal module, and does not have a stable

14+-- API or name. Functions in this module may not check or enforce

15+-- preconditions expected by public modules. Use at your own risk!

16+--

17+-- Size hints.

18+

19+module Data.Text.Internal.Fusion.Size

20+ (

21+ Size

22+ , exactly

23+ , exactSize

24+ , maxSize

25+ , betweenSize

26+ , unknownSize

27+ , smaller

28+ , larger

29+ , upperBound

30+ , lowerBound

31+ , compareSize

32+ , isEmpty

33+ ) where

34+

35+import Data.Text.Internal (mul)

36+#if defined(ASSERTS)

37+import Control.Exception (assert)

38+#endif

39+

40+data Size = Between {-# UNPACK #-} !Int {-# UNPACK #-} !Int -- ^ Lower and upper bounds on size.

41+ | Unknown -- ^ Unknown size.

42+ deriving (Eq, Show)

43+

44+exactly :: Size -> Maybe Int

45+exactly (Between na nb) | na == nb = Just na

46+exactly _ = Nothing

47+{-# INLINE exactly #-}

48+

49+exactSize :: Int -> Size

50+exactSize n =

51+#if defined(ASSERTS)

52+ assert (n >= 0)

53+#endif

54+ Between n n

55+{-# INLINE exactSize #-}

56+

57+maxSize :: Int -> Size

58+maxSize n =

59+#if defined(ASSERTS)

60+ assert (n >= 0)

61+#endif

62+ Between 0 n

63+{-# INLINE maxSize #-}

64+

65+betweenSize :: Int -> Int -> Size

66+betweenSize m n =

67+#if defined(ASSERTS)

68+ assert (m >= 0)

69+ assert (n >= m)

70+#endif

71+ Between m n

72+{-# INLINE betweenSize #-}

73+

74+unknownSize :: Size

75+unknownSize = Unknown

76+{-# INLINE unknownSize #-}

77+

78+instance Num Size where

79+ (+) = addSize

80+ (-) = subtractSize

81+ (*) = mulSize

82+

83+ fromInteger = f where f = exactSize . fromInteger

84+ {-# INLINE f #-}

85+

86+add :: Int -> Int -> Int

87+add m n | mn >= 0 = mn

88+ | otherwise = overflowError

89+ where mn = m + n

90+{-# INLINE add #-}

91+

92+addSize :: Size -> Size -> Size

93+addSize (Between ma mb) (Between na nb) = Between (add ma na) (add mb nb)

94+addSize _ _ = Unknown

95+{-# INLINE addSize #-}

96+

97+subtractSize :: Size -> Size -> Size

98+subtractSize (Between ma mb) (Between na nb) = Between (max (ma-nb) 0) (max (mb-na) 0)

99+subtractSize a@(Between 0 _) Unknown = a

100+subtractSize (Between _ mb) Unknown = Between 0 mb

101+subtractSize _ _ = Unknown

102+{-# INLINE subtractSize #-}

103+

104+mulSize :: Size -> Size -> Size

105+mulSize (Between ma mb) (Between na nb) = Between (mul ma na) (mul mb nb)

106+mulSize _ _ = Unknown

107+{-# INLINE mulSize #-}

108+

109+-- | Minimum of two size hints.

110+smaller :: Size -> Size -> Size

111+smaller a@(Between ma mb) b@(Between na nb)

112+ | mb <= na = a

113+ | nb <= ma = b

114+ | otherwise = Between (ma `min` na) (mb `min` nb)

115+smaller a@(Between 0 _) Unknown = a

116+smaller (Between _ mb) Unknown = Between 0 mb

117+smaller Unknown b@(Between 0 _) = b

118+smaller Unknown (Between _ nb) = Between 0 nb

119+smaller Unknown Unknown = Unknown

120+{-# INLINE smaller #-}

121+

122+-- | Maximum of two size hints.

123+larger :: Size -> Size -> Size

124+larger a@(Between ma mb) b@(Between na nb)

125+ | ma >= nb = a

126+ | na >= mb = b

127+ | otherwise = Between (ma `max` na) (mb `max` nb)

128+larger _ _ = Unknown

129+{-# INLINE larger #-}

130+

131+-- | Compute the maximum size from a size hint, if possible.

132+upperBound :: Int -> Size -> Int

133+upperBound _ (Between _ n) = n

134+upperBound k _ = k

135+{-# INLINE upperBound #-}

136+

137+-- | Compute the maximum size from a size hint, if possible.

138+lowerBound :: Int -> Size -> Int

139+lowerBound _ (Between n _) = n

140+lowerBound k _ = k

141+{-# INLINE lowerBound #-}

142+

143+compareSize :: Size -> Int -> Maybe Ordering

144+compareSize (Between ma mb) n

145+ | mb < n = Just LT

146+ | ma > n = Just GT

147+ | ma == n && mb == n = Just EQ

148+compareSize _ _ = Nothing

149+

150+

151+isEmpty :: Size -> Bool

152+isEmpty (Between _ n) = n <= 0

153+isEmpty _ = False

154+{-# INLINE isEmpty #-}

155+

156+overflowError :: Int

157+overflowError = error "Data.Text.Internal.Fusion.Size: size overflow"

Data/Text/Fusion/Internal.hs → Data/Text/Internal/Fusion/Types.hsrenamed+16−18

@@ -1,58 +1,56 @@

11{-# LANGUAGE BangPatterns, ExistentialQuantification #-}

22-- |

3−-- Module : Data.Text.Fusion.Internal

3+-- Module : Data.Text.Internal.Fusion.Types

44-- Copyright : (c) Tom Harper 2008-2009,

55-- (c) Bryan O'Sullivan 2009,

66-- (c) Duncan Coutts 2009,

77-- (c) Jasper Van der Jeugt 2011

88--

99-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

10+-- Maintainer : bos@serpentine.com

1211-- Stability : experimental

1312-- Portability : GHC

1413--

14+-- /Warning/: this is an internal module, and does not have a stable

15+-- API or name. Functions in this module may not check or enforce

16+-- preconditions expected by public modules. Use at your own risk!

17+--

1518-- Core stream fusion functionality for text.

1619

17−module Data.Text.Fusion.Internal

20+module Data.Text.Internal.Fusion.Types

1821 (

1922 CC(..)

20− , M(..)

21− , M8

2223 , PairS(..)

24+ , Scan(..)

2325 , RS(..)

2426 , Step(..)

2527 , Stream(..)

26− , Switch(..)

2728 , empty

2829 ) where

2930

30−import Data.Text.Fusion.Size

31+import Data.Text.Internal.Fusion.Size

3132import Data.Word (Word8)

3233

3334-- | Specialised tuple for case conversion.

3435data CC s = CC !s {-# UNPACK #-} !Char {-# UNPACK #-} !Char

3536

36−-- | Specialised, strict Maybe-like type.

37−data M a = N

38− | J !a

39−

40−type M8 = M Word8

41−

42−-- Restreaming state.

37+-- | Restreaming state.

4338data RS s

4439 = RS0 !s

4540 | RS1 !s {-# UNPACK #-} !Word8

4641 | RS2 !s {-# UNPACK #-} !Word8 {-# UNPACK #-} !Word8

4742 | RS3 !s {-# UNPACK #-} !Word8 {-# UNPACK #-} !Word8 {-# UNPACK #-} !Word8

4843

49−infixl 2 :*:

44+-- | Strict pair.

5045data PairS a b = !a :*: !b

5146 -- deriving (Eq, Ord, Show)

47+infixl 2 :*:

5248

53−-- | Allow a function over a stream to switch between two states.

54−data Switch = S1 | S2

49+-- | An intermediate result in a scan.

50+data Scan s = Scan1 {-# UNPACK #-} !Char !s

51+ | Scan2 {-# UNPACK #-} !Char !s

5552

53+-- | Intermediate result in a processing pipeline.

5654data Step s a = Done

5755 | Skip !s

5856 | Yield !a !s

Data/Text/IO/Internal.hs → Data/Text/Internal/IO.hsrenamed+9−5

@@ -1,6 +1,6 @@

11{-# LANGUAGE BangPatterns, CPP, RecordWildCards #-}

22-- |

3−-- Module : Data.Text.IO.Internal

3+-- Module : Data.Text.Internal.IO

66-- License : BSD-style

@@ -8,9 +8,13 @@

88-- Stability : experimental

99-- Portability : GHC

1010--

11+-- /Warning/: this is an internal module, and does not have a stable

12+-- API or name. Functions in this module may not check or enforce

13+-- preconditions expected by public modules. Use at your own risk!

14+--

1115-- Low-level support for text I\/O.

1216

13−module Data.Text.IO.Internal

17+module Data.Text.Internal.IO

1418 (

1519 hGetLineWith

1620 , readChunk

@@ -19,9 +23,9 @@ module Data.Text.IO.Internal

1923import qualified Control.Exception as E

2024import Data.IORef (readIORef, writeIORef)

2125import Data.Text (Text)

22−import Data.Text.Fusion (unstream)

23−import Data.Text.Fusion.Internal (Step(..), Stream(..))

24−import Data.Text.Fusion.Size (exactSize, maxSize)

26+import Data.Text.Internal.Fusion (unstream)

27+import Data.Text.Internal.Fusion.Types (Step(..), Stream(..))

28+import Data.Text.Internal.Fusion.Size (exactSize, maxSize)

2529import Data.Text.Unsafe (inlinePerformIO)

2630import Foreign.Storable (peekElemOff)

2731import GHC.IO.Buffer (Buffer(..), CharBuffer, RawCharBuffer, bufferAdjustL,

Data/Text/Internal/Lazy.hsnew file+119

@@ -0,0 +1,119 @@

1+{-# LANGUAGE BangPatterns, DeriveDataTypeable #-}

2+{-# OPTIONS_HADDOCK not-home #-}

4+-- |

5+-- Module : Data.Text.Internal.Lazy

7+--

8+-- License : BSD-style

9+-- Maintainer : bos@serpentine.com

10+-- Stability : experimental

11+-- Portability : GHC

12+--

13+-- /Warning/: this is an internal module, and does not have a stable

14+-- API or name. Functions in this module may not check or enforce

15+-- preconditions expected by public modules. Use at your own risk!

16+--

17+-- A module containing private 'Text' internals. This exposes the

18+-- 'Text' representation and low level construction functions.

19+-- Modules which extend the 'Text' system may need to use this module.

20+

21+module Data.Text.Internal.Lazy

22+ (

23+ Text(..)

24+ , chunk

25+ , empty

26+ , foldrChunks

27+ , foldlChunks

28+ -- * Data type invariant and abstraction functions

29+

30+ -- $invariant

31+ , strictInvariant

32+ , lazyInvariant

33+ , showStructure

34+

35+ -- * Chunk allocation sizes

36+ , defaultChunkSize

37+ , smallChunkSize

38+ , chunkOverhead

39+ ) where

40+

41+import Data.Text ()

42+import Data.Text.Internal.Unsafe.Shift (shiftL)

43+import Data.Typeable (Typeable)

44+import Foreign.Storable (sizeOf)

45+import qualified Data.Text.Internal as T

46+

47+data Text = Empty

48+ | Chunk {-# UNPACK #-} !T.Text Text

49+ deriving (Typeable)

50+

51+-- $invariant

52+--

53+-- The data type invariant for lazy 'Text': Every 'Text' is either 'Empty' or

54+-- consists of non-null 'T.Text's. All functions must preserve this,

55+-- and the QC properties must check this.

56+

57+-- | Check the invariant strictly.

58+strictInvariant :: Text -> Bool

59+strictInvariant Empty = True

60+strictInvariant x@(Chunk (T.Text _ _ len) cs)

61+ | len > 0 = strictInvariant cs

62+ | otherwise = error $ "Data.Text.Lazy: invariant violation: "

63+ ++ showStructure x

64+

65+-- | Check the invariant lazily.

66+lazyInvariant :: Text -> Text

67+lazyInvariant Empty = Empty

68+lazyInvariant x@(Chunk c@(T.Text _ _ len) cs)

69+ | len > 0 = Chunk c (lazyInvariant cs)

70+ | otherwise = error $ "Data.Text.Lazy: invariant violation: "

71+ ++ showStructure x

72+

73+-- | Display the internal structure of a lazy 'Text'.

74+showStructure :: Text -> String

75+showStructure Empty = "Empty"

76+showStructure (Chunk t Empty) = "Chunk " ++ show t ++ " Empty"

77+showStructure (Chunk t ts) =

78+ "Chunk " ++ show t ++ " (" ++ showStructure ts ++ ")"

79+

80+-- | Smart constructor for 'Chunk'. Guarantees the data type invariant.

81+chunk :: T.Text -> Text -> Text

82+{-# INLINE chunk #-}

83+chunk t@(T.Text _ _ len) ts | len == 0 = ts

84+ | otherwise = Chunk t ts

85+

86+-- | Smart constructor for 'Empty'.

87+empty :: Text

88+{-# INLINE [0] empty #-}

89+empty = Empty

90+

91+-- | Consume the chunks of a lazy 'Text' with a natural right fold.

92+foldrChunks :: (T.Text -> a -> a) -> a -> Text -> a

93+foldrChunks f z = go

94+ where go Empty = z

95+ go (Chunk c cs) = f c (go cs)

96+{-# INLINE foldrChunks #-}

97+

98+-- | Consume the chunks of a lazy 'Text' with a strict, tail-recursive,

99+-- accumulating left fold.

100+foldlChunks :: (a -> T.Text -> a) -> a -> Text -> a

101+foldlChunks f z = go z

102+ where go !a Empty = a

103+ go !a (Chunk c cs) = go (f a c) cs

104+{-# INLINE foldlChunks #-}

105+

106+-- | Currently set to 16 KiB, less the memory management overhead.

107+defaultChunkSize :: Int

108+defaultChunkSize = 16384 - chunkOverhead

109+{-# INLINE defaultChunkSize #-}

110+

111+-- | Currently set to 128 bytes, less the memory management overhead.

112+smallChunkSize :: Int

113+smallChunkSize = 128 - chunkOverhead

114+{-# INLINE smallChunkSize #-}

115+

116+-- | The memory management overhead. Currently this is tuned for GHC only.

117+chunkOverhead :: Int

118+chunkOverhead = sizeOf (undefined :: Int) `shiftL` 1

119+{-# INLINE chunkOverhead #-}

Data/Text/Lazy/Encoding/Fusion.hs → Data/Text/Internal/Lazy/Encoding/Fusion.hsrenamed+15−12

@@ -5,15 +5,18 @@

66--

77-- License : BSD-style

8−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

9−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

109-- Stability : experimental

1110-- Portability : portable

1211--

12+-- /Warning/: this is an internal module, and does not have a stable

13+-- API or name. Functions in this module may not check or enforce

14+-- preconditions expected by public modules. Use at your own risk!

15+--

1316-- Fusible 'Stream'-oriented functions for converting between lazy

1417-- 'Text' and several common encodings.

1518

16−module Data.Text.Lazy.Encoding.Fusion

19+module Data.Text.Internal.Lazy.Encoding.Fusion

1720 (

1821 -- * Streaming

1922 -- streamASCII

@@ -26,22 +29,22 @@ module Data.Text.Lazy.Encoding.Fusion

2629 -- * Unstreaming

2730 , unstream

2831

29− , module Data.Text.Encoding.Fusion.Common

32+ , module Data.Text.Internal.Encoding.Fusion.Common

3033 ) where

3134

3235import Data.ByteString.Lazy.Internal (ByteString(..), defaultChunkSize)

3336import qualified Data.ByteString as B

3437import qualified Data.ByteString.Unsafe as B

35−import Data.Text.Encoding.Fusion.Common

38+import Data.Text.Internal.Encoding.Fusion.Common

3639import Data.Text.Encoding.Error

37−import Data.Text.Fusion (Step(..), Stream(..))

38−import Data.Text.Fusion.Size

39−import Data.Text.UnsafeChar (unsafeChr, unsafeChr8, unsafeChr32)

40−import Data.Text.UnsafeShift (shiftL)

40+import Data.Text.Internal.Fusion (Step(..), Stream(..))

41+import Data.Text.Internal.Fusion.Size

42+import Data.Text.Internal.Unsafe.Char (unsafeChr, unsafeChr8, unsafeChr32)

43+import Data.Text.Internal.Unsafe.Shift (shiftL)

4144import Data.Word (Word8, Word16, Word32)

42−import qualified Data.Text.Encoding.Utf8 as U8

43−import qualified Data.Text.Encoding.Utf16 as U16

44−import qualified Data.Text.Encoding.Utf32 as U32

45+import qualified Data.Text.Internal.Encoding.Utf8 as U8

46+import qualified Data.Text.Internal.Encoding.Utf16 as U16

47+import qualified Data.Text.Internal.Encoding.Utf32 as U32

4548import Data.Text.Unsafe (unsafeDupablePerformIO)

4649import Foreign.ForeignPtr (withForeignPtr, ForeignPtr)

4750import Foreign.Storable (pokeByteOff)

Data/Text/Lazy/Fusion.hs → Data/Text/Internal/Lazy/Fusion.hsrenamed+12−9

@@ -4,14 +4,17 @@

55--

66-- License : BSD-style

7−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

8−-- duncan@haskell.org

7+-- Maintainer : bos@serpentine.com

98-- Stability : experimental

109-- Portability : GHC

1110--

11+-- /Warning/: this is an internal module, and does not have a stable

12+-- API or name. Functions in this module may not check or enforce

13+-- preconditions expected by public modules. Use at your own risk!

14+--

1215-- Core stream fusion functionality for text.

1316

14−module Data.Text.Lazy.Fusion

17+module Data.Text.Internal.Lazy.Fusion

1518 (

1619 stream

1720 , unstream

@@ -23,15 +26,15 @@ module Data.Text.Lazy.Fusion

2326 ) where

2427

2528import Prelude hiding (length)

26−import qualified Data.Text.Fusion.Common as S

29+import qualified Data.Text.Internal.Fusion.Common as S

2730import Control.Monad.ST (runST)

28−import Data.Text.Fusion.Internal

29−import Data.Text.Fusion.Size (isEmpty, unknownSize)

30−import Data.Text.Lazy.Internal

31+import Data.Text.Internal.Fusion.Types

32+import Data.Text.Internal.Fusion.Size (isEmpty, unknownSize)

33+import Data.Text.Internal.Lazy

3134import qualified Data.Text.Internal as I

3235import qualified Data.Text.Array as A

33−import Data.Text.UnsafeChar (unsafeWrite)

34−import Data.Text.UnsafeShift (shiftL)

36+import Data.Text.Internal.Unsafe.Char (unsafeWrite)

37+import Data.Text.Internal.Unsafe.Shift (shiftL)

3538import Data.Text.Unsafe (Iter(..), iter)

3639import Data.Int (Int64)

3740

Data/Text/Lazy/Search.hs → Data/Text/Internal/Lazy/Search.hsrenamed+9−6

@@ -5,16 +5,19 @@

66--

77-- License : BSD-style

8−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

9−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

109-- Stability : experimental

1110-- Portability : GHC

1211--

12+-- /Warning/: this is an internal module, and does not have a stable

13+-- API or name. Functions in this module may not check or enforce

14+-- preconditions expected by public modules. Use at your own risk!

15+--

1316-- Fast substring search for lazy 'Text', based on work by Boyer,

1417-- Moore, Horspool, Sunday, and Lundh. Adapted from the strict

1518-- implementation.

1619

17−module Data.Text.Lazy.Search

20+module Data.Text.Internal.Lazy.Search

1821 (

1922 indices

2023 ) where

@@ -23,10 +26,10 @@ import qualified Data.Text.Array as A

2326import Data.Int (Int64)

2427import Data.Word (Word16, Word64)

2528import qualified Data.Text.Internal as T

26−import Data.Text.Fusion.Internal (PairS(..))

27−import Data.Text.Lazy.Internal (Text(..), foldlChunks)

29+import Data.Text.Internal.Fusion.Types (PairS(..))

30+import Data.Text.Internal.Lazy (Text(..), foldlChunks)

2831import Data.Bits ((.|.), (.&.))

29−import Data.Text.UnsafeShift (shiftL)

32+import Data.Text.Internal.Unsafe.Shift (shiftL)

3033

3134-- | /O(n+m)/ Find the offsets of all non-overlapping indices of

3235-- @needle@ within @haystack@.

Data/Text/Private.hs → Data/Text/Internal/Private.hsrenamed+6−6

@@ -1,7 +1,7 @@

11{-# LANGUAGE BangPatterns, Rank2Types, UnboxedTuples #-}

33-- |

4−-- Module : Data.Text.Private

4+-- Module : Data.Text.Internal.Private

66--

77-- License : BSD-style

@@ -9,21 +9,21 @@

99-- Stability : experimental

1010-- Portability : GHC

1111

12−module Data.Text.Private

12+module Data.Text.Internal.Private

1313 (

1414 runText

1515 , span_

1616 ) where

1717

1818import Control.Monad.ST (ST, runST)

19−import Data.Text.Internal (Text(..), textP)

19+import Data.Text.Internal (Text(..), text)

2020import Data.Text.Unsafe (Iter(..), iter)

2121import qualified Data.Text.Array as A

2222

2323span_ :: (Char -> Bool) -> Text -> (# Text, Text #)

2424span_ p t@(Text arr off len) = (# hd,tl #)

25− where hd = textP arr off k

26− tl = textP arr (off+k) (len-k)

25+ where hd = text arr off k

26+ tl = text arr (off+k) (len-k)

2727 !k = loop 0

2828 loop !i | i < len && p c = loop (i+d)

2929 | otherwise = i

@@ -33,5 +33,5 @@ span_ p t@(Text arr off len) = (# hd,tl #)

3333runText :: (forall s. (A.MArray s -> Int -> ST s Text) -> ST s Text) -> Text

3434runText act = runST (act $ \ !marr !len -> do

3535 arr <- A.unsafeFreeze marr

36− return $! textP arr 0 len)

36+ return $! text arr 0 len)

3737{-# INLINE runText #-}

Data/Text/Internal/Read.hsnew file+62

@@ -0,0 +1,62 @@

1+-- |

2+-- Module : Data.Text.Internal.Read

4+--

5+-- License : BSD-style

6+-- Maintainer : bos@serpentine.com

7+-- Stability : experimental

8+-- Portability : GHC

9+--

10+-- Common internal functions for reading textual data.

11+module Data.Text.Internal.Read

12+ (

13+ IReader

14+ , IParser(..)

15+ , T(..)

16+ , digitToInt

17+ , hexDigitToInt

18+ , perhaps

19+ ) where

20+

21+import Control.Applicative as App (Applicative(..))

22+import Control.Arrow (first)

23+import Control.Monad (ap)

24+import Data.Char (ord)

25+

26+type IReader t a = t -> Either String (a,t)

27+

28+newtype IParser t a = P {

29+ runP :: IReader t a

30+ }

31+

32+instance Functor (IParser t) where

33+ fmap f m = P $ fmap (first f) . runP m

34+

35+instance Applicative (IParser t) where

36+ pure a = P $ \t -> Right (a,t)

37+ {-# INLINE pure #-}

38+ (<*>) = ap

39+

40+instance Monad (IParser t) where

41+ return = App.pure

42+ m >>= k = P $ \t -> case runP m t of

43+ Left err -> Left err

44+ Right (a,t') -> runP (k a) t'

45+ {-# INLINE (>>=) #-}

46+ fail msg = P $ \_ -> Left msg

47+

48+data T = T !Integer !Int

49+

50+perhaps :: a -> IParser t a -> IParser t a

51+perhaps def m = P $ \t -> case runP m t of

52+ Left _ -> Right (def,t)

53+ r@(Right _) -> r

54+

55+hexDigitToInt :: Char -> Int

56+hexDigitToInt c

57+ | c >= '0' && c <= '9' = ord c - ord '0'

58+ | c >= 'a' && c <= 'f' = ord c - (ord 'a' - 10)

59+ | otherwise = ord c - (ord 'A' - 10)

60+

61+digitToInt :: Char -> Int

62+digitToInt c = ord c - ord '0'

Data/Text/Search.hs → Data/Text/Internal/Search.hsrenamed+4−5

@@ -1,12 +1,11 @@

11{-# LANGUAGE BangPatterns, ScopedTypeVariables #-}

33-- |

4−-- Module : Data.Text.Search

4+-- Module : Data.Text.Internal.Search

55-- Copyright : (c) Bryan O'Sullivan 2009

66--

77-- License : BSD-style

8−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

9−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

109-- Stability : experimental

1110-- Portability : GHC

1211--

@@ -27,7 +26,7 @@

2726-- * F. Lundh: The Fast Search Algorithm.

2827-- <http://effbot.org/zone/stringlib.htm> (2006)

2928

30−module Data.Text.Search

29+module Data.Text.Internal.Search

3130 (

3231 indices

3332 ) where

@@ -36,7 +35,7 @@ import qualified Data.Text.Array as A

3635import Data.Word (Word64)

3736import Data.Text.Internal (Text(..))

3837import Data.Bits ((.|.), (.&.))

39−import Data.Text.UnsafeShift (shiftL)

38+import Data.Text.Internal.Unsafe.Shift (shiftL)

4039

4140data T = {-# UNPACK #-} !Word64 :* {-# UNPACK #-} !Int

4241

Data/Text/Unsafe/Base.hs → Data/Text/Internal/Unsafe.hsrenamed+11−6

@@ -1,16 +1,21 @@

11{-# LANGUAGE CPP, MagicHash, UnboxedTuples #-}

2+{-# OPTIONS_HADDOCK not-home #-}

24-- |

3−-- Module : Data.Text.Unsafe.Base

5+-- Module : Data.Text.Internal.Unsafe

57-- License : BSD-style

6−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

7−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

89-- Stability : experimental

910-- Portability : portable

1011--

11−-- A module containing unsafe operations, for very very careful use in

12−-- heavily tested code.

13−module Data.Text.Unsafe.Base

12+-- /Warning/: this is an internal module, and does not have a stable

13+-- API or name. Functions in this module may not check or enforce

14+-- preconditions expected by public modules. Use at your own risk!

15+--

16+-- A module containing /unsafe/ operations, for /very very careful/ use

17+-- in /heavily tested/ code.

18+module Data.Text.Internal.Unsafe

1419 (

1520 inlineInterleaveST

1621 , inlinePerformIO

Data/Text/UnsafeChar.hs → Data/Text/Internal/Unsafe/Char.hsrenamed+8−5

@@ -1,19 +1,22 @@

11{-# LANGUAGE CPP, MagicHash #-}

33-- |

4−-- Module : Data.Text.UnsafeChar

4+-- Module : Data.Text.Internal.Unsafe.Char

88--

99-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

10+-- Maintainer : bos@serpentine.com

1211-- Stability : experimental

1312-- Portability : GHC

1413--

14+-- /Warning/: this is an internal module, and does not have a stable

15+-- API or name. Functions in this module may not check or enforce

16+-- preconditions expected by public modules. Use at your own risk!

17+--

1518-- Fast character manipulation functions.

16−module Data.Text.UnsafeChar

19+module Data.Text.Internal.Unsafe.Char

1720 (

1821 ord

1922 , unsafeChr

@@ -28,7 +31,7 @@ import Control.Exception (assert)

2831#endif

2932import Control.Monad.ST (ST)

3033import Data.Bits ((.&.))

31−import Data.Text.UnsafeShift (shiftR)

34+import Data.Text.Internal.Unsafe.Shift (shiftR)

3235import GHC.Exts (Char(..), Int(..), chr#, ord#, word2Int#)

3336import GHC.Word (Word8(..), Word16(..), Word32(..))

3437import qualified Data.Text.Array as A

Data/Text/UnsafeShift.hs → Data/Text/Internal/Unsafe/Shift.hsrenamed+7−4

@@ -1,18 +1,21 @@

11{-# LANGUAGE MagicHash #-}

33-- |

4−-- Module : Data.Text.UnsafeShift

4+-- Module : Data.Text.Internal.Unsafe.Shift

55-- Copyright : (c) Bryan O'Sullivan 2009

66--

77-- License : BSD-style

8−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

9−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

109-- Stability : experimental

1110-- Portability : GHC

1211--

12+-- /Warning/: this is an internal module, and does not have a stable

13+-- API or name. Functions in this module may not check or enforce

14+-- preconditions expected by public modules. Use at your own risk!

15+--

1316-- Fast, unchecked bit shifting functions.

1417

15−module Data.Text.UnsafeShift

18+module Data.Text.Internal.Unsafe.Shift

1619 (

1720 UnsafeShift(..)

1821 ) where

Data/Text/Lazy.hs+209−35

@@ -3,13 +3,16 @@

33#if __GLASGOW_HASKELL__ >= 702

44{-# LANGUAGE Trustworthy #-}

55#endif

6+#if __GLASGOW_HASKELL__ >= 708

7+{-# LANGUAGE TypeFamilies #-}

8+#endif

610-- |

711-- Module : Data.Text.Lazy

913--

1014-- License : BSD-style

11−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

12−-- duncan@haskell.org

15+-- Maintainer : bos@serpentine.com

1316-- Stability : experimental

1417-- Portability : GHC

1518--

@@ -122,7 +125,10 @@ module Data.Text.Lazy

122125 , mapAccumR

123126

124127 -- ** Generation and unfolding

128+ , repeat

125129 , replicate

130+ , cycle

131+ , iterate

126132 , unfoldr

127133 , unfoldrN

128134

@@ -130,8 +136,11 @@ module Data.Text.Lazy

130136

131137 -- ** Breaking strings

132138 , take

139+ , takeEnd

133140 , drop

141+ , dropEnd

134142 , takeWhile

143+ , takeWhileEnd

135144 , dropWhile

136145 , dropWhileEnd

137146 , dropAround

@@ -202,27 +211,39 @@ import Control.DeepSeq (NFData(..))

202211import Data.Int (Int64)

203212import qualified Data.List as L

204213import Data.Char (isSpace)

205−import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf))

206−import Data.Data (mkNoRepType)

214+import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,

215+ Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))

216+import Data.Binary (Binary(get, put))

207217import Data.Monoid (Monoid(..))

218+#if MIN_VERSION_base(4,9,0)

219+import Data.Semigroup (Semigroup(..))

220+#endif

208221import Data.String (IsString(..))

209222import qualified Data.Text as T

210223import qualified Data.Text.Internal as T

211−import qualified Data.Text.Fusion.Common as S

224+import qualified Data.Text.Internal.Fusion.Common as S

212225import qualified Data.Text.Unsafe as T

213−import qualified Data.Text.Lazy.Fusion as S

214−import Data.Text.Fusion.Internal (PairS(..))

215−import Data.Text.Lazy.Fusion (stream, unstream)

216−import Data.Text.Lazy.Internal (Text(..), chunk, empty, foldlChunks, foldrChunks)

217−import Data.Text.Internal (firstf, safe, textP)

218−import qualified Data.Text.Util as U

219−import Data.Text.Lazy.Search (indices)

226+import qualified Data.Text.Internal.Lazy.Fusion as S

227+import Data.Text.Internal.Fusion.Types (PairS(..))

228+import Data.Text.Internal.Lazy.Fusion (stream, unstream)

229+import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldlChunks,

230+ foldrChunks, smallChunkSize)

231+import Data.Text.Internal (firstf, safe, text)

232+import Data.Text.Lazy.Encoding (decodeUtf8', encodeUtf8)

233+import qualified Data.Text.Internal.Functions as F

234+import Data.Text.Internal.Lazy.Search (indices)

220235#if __GLASGOW_HASKELL__ >= 702

221236import qualified GHC.CString as GHC

222237#else

223238import qualified GHC.Base as GHC

224239#endif

240+#if __GLASGOW_HASKELL__ >= 708

241+import qualified GHC.Exts as Exts

242+#endif

225243import GHC.Prim (Addr#)

244+#if MIN_VERSION_base(4,7,0)

245+import Text.Printf (PrintfArg, formatArg, formatString)

246+#endif

226247

227248-- $fusion

228249--

@@ -320,25 +341,71 @@ instance Show Text where

320341instance Read Text where

321342 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]

322343

344+#if MIN_VERSION_base(4,9,0)

345+-- Semigroup orphan instances for older GHCs are provided by

346+-- 'semigroups` package

347+

348+instance Semigroup Text where

349+ (<>) = append

350+#endif

351+

323352instance Monoid Text where

324353 mempty = empty

354+#if MIN_VERSION_base(4,9,0)

355+ mappend = (<>) -- future-proof definition

356+#else

325357 mappend = append

358+#endif

326359 mconcat = concat

327360

328361instance IsString Text where

329362 fromString = pack

330363

364+#if __GLASGOW_HASKELL__ >= 708

365+instance Exts.IsList Text where

366+ type Item Text = Char

367+ fromList = pack

368+ toList = unpack

369+#endif

370+

331371#if defined(HAVE_DEEPSEQ)

332372instance NFData Text where

333373 rnf Empty = ()

334374 rnf (Chunk _ ts) = rnf ts

335375#endif

336376

377+instance Binary Text where

378+ put t = put (encodeUtf8 t)

379+ get = do

380+ bs <- get

381+ case decodeUtf8' bs of

382+ P.Left exn -> P.fail (P.show exn)

383+ P.Right a -> P.return a

384+

385+-- | This instance preserves data abstraction at the cost of inefficiency.

386+-- We omit reflection services for the sake of data abstraction.

387+--

388+-- This instance was created by copying the updated behavior of

389+-- @"Data.Text".@'Data.Text.Text'

337390instance Data Text where

338391 gfoldl f z txt = z pack `f` (unpack txt)

339− toConstr _ = error "Data.Text.Lazy.Text.toConstr"

340− gunfold _ _ = error "Data.Text.Lazy.Text.gunfold"

341− dataTypeOf _ = mkNoRepType "Data.Text.Lazy.Text"

392+ toConstr _ = packConstr

393+ gunfold k z c = case constrIndex c of

394+ 1 -> k (z pack)

395+ _ -> error "Data.Text.Lazy.Text.gunfold"

396+ dataTypeOf _ = textDataType

397+

398+#if MIN_VERSION_base(4,7,0)

399+-- | Only defined for @base-4.7.0.0@ and later

400+instance PrintfArg Text where

401+ formatArg txt = formatString $ unpack txt

402+#endif

403+

404+packConstr :: Constr

405+packConstr = mkConstr textDataType "pack" [] Prefix

406+

407+textDataType :: DataType

408+textDataType = mkDataType "Data.Text.Lazy.Text" [packConstr]

342409

343410-- | /O(n)/ Convert a 'String' into a 'Text'.

344411--

@@ -366,6 +433,14 @@ unpackCString# addr# = unstream (S.streamCString# addr#)

366433 unstream (S.streamList (L.map safe (GHC.unpackCStringUtf8# a)))

367434 = unpackCString# a #-}

368435

436+{-# RULES "LAZY TEXT empty literal"

437+ unstream (S.streamList (L.map safe []))

438+ = Empty #-}

439+

440+{-# RULES "LAZY TEXT empty literal" forall a.

441+ unstream (S.streamList (L.map safe [a]))

442+ = Chunk (T.singleton a) Empty #-}

443+

369444-- | /O(1)/ Convert a character into a Text. Subject to fusion.

370445-- Performs replacement on invalid scalar values.

371446singleton :: Char -> Text

@@ -470,7 +545,7 @@ tail Empty = emptyError "tail"

470545 unstream (S.tail (stream t)) = tail t

471546 #-}

472547

473−-- | /O(1)/ Returns all but the last character of a 'Text', which must

548+-- | /O(n\/c)/ Returns all but the last character of a 'Text', which must

474549-- be non-empty. Subject to fusion.

475550init :: Text -> Text

476551init (Chunk t0 ts0) = go t0 ts0

@@ -506,7 +581,7 @@ isSingleton :: Text -> Bool

506581isSingleton = S.isSingleton . stream

507582{-# INLINE isSingleton #-}

508583

509−-- | /O(1)/ Returns the last character of a 'Text', which must be

584+-- | /O(n\/c)/ Returns the last character of a 'Text', which must be

510585-- non-empty. Subject to fusion.

511586last :: Text -> Char

512587last Empty = emptyError "last"

@@ -561,7 +636,7 @@ map f t = unstream (S.map (safe . f) (stream t))

561636-- 'Text's and concatenates the list after interspersing the first

562637-- argument between each element of the list.

563638intercalate :: Text -> [Text] -> Text

564−intercalate t = concat . (U.intersperse t)

639+intercalate t = concat . (F.intersperse t)

565640{-# INLINE intercalate #-}

566641

567642-- | /O(n)/ The 'intersperse' function takes a character and places it

@@ -640,13 +715,36 @@ reverse = rev Empty

640715 where rev a Empty = a

641716 rev a (Chunk t ts) = rev (Chunk (T.reverse t) a) ts

642717

643−-- | /O(m+n)/ Replace every occurrence of one substring with another.

718+-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in

719+-- @haystack@ with @replacement@.

720+--

721+-- This function behaves as though it was defined as follows:

722+--

723+-- @

724+-- replace needle replacement haystack =

725+-- 'intercalate' replacement ('splitOn' needle haystack)

726+-- @

727+--

728+-- As this suggests, each occurrence is replaced exactly once. So if

729+-- @needle@ occurs in @replacement@, that occurrence will /not/ itself

730+-- be replaced recursively:

731+--

732+-- > replace "oo" "foo" "oo" == "foo"

733+--

734+-- In cases where several instances of @needle@ overlap, only the

735+-- first one will be replaced:

736+--

737+-- > replace "ofo" "bar" "ofofo" == "barfo"

644738--

645739-- In (unlikely) bad cases, this function's time complexity degrades

646740-- towards /O(n*m)/.

647−replace :: Text -- ^ Text to search for

648− -> Text -- ^ Replacement text

649− -> Text -- ^ Input text

741+replace :: Text

742+ -- ^ @needle@ to search for. If this string is empty, an

743+ -- error will occur.

744+ -> Text

745+ -- ^ @replacement@ to replace @needle@ with.

746+ -> Text

747+ -- ^ @haystack@ in which to search.

650748 -> Text

651749replace s d = intercalate d . splitOn s

652750{-# INLINE replace #-}

@@ -783,13 +881,13 @@ concatMap f = concat . foldr ((:) . f) []

783881{-# INLINE concatMap #-}

784882

785883-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the

786−-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.

884+-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.

787885any :: (Char -> Bool) -> Text -> Bool

788886any p t = S.any p (stream t)

789887{-# INLINE any #-}

790888

791889-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the

792−-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.

890+-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.

793891all :: (Char -> Bool) -> Text -> Bool

794892all p t = S.all p (stream t)

795893{-# INLINE all #-}

@@ -872,6 +970,12 @@ mapAccumR f = go

872970 go z Empty = (z, Empty)

873971{-# INLINE mapAccumR #-}

874972

973+-- | @'repeat' x@ is an infinite 'Text', with @x@ the value of every

974+-- element.

975+repeat :: Char -> Text

976+repeat c = let t = Chunk (T.replicate smallChunkSize (T.singleton c)) t

977+ in t

978+

875979-- | /O(n*m)/ 'replicate' @n@ @t@ is a 'Text' consisting of the input

876980-- @t@ repeated @n@ times.

877981replicate :: Int64 -> Text -> Text

@@ -881,7 +985,22 @@ replicate n t

881985 | otherwise = concat (rep 0)

882986 where rep !i | i >= n = []

883987 | otherwise = t : rep (i+1)

884−{-# INLINE replicate #-}

988+{-# INLINE [1] replicate #-}

989+

990+-- | 'cycle' ties a finite, non-empty 'Text' into a circular one, or

991+-- equivalently, the infinite repetition of the original 'Text'.

992+cycle :: Text -> Text

993+cycle Empty = emptyError "cycle"

994+cycle t = let t' = foldrChunks Chunk t' t

995+ in t'

996+

997+-- | @'iterate' f x@ returns an infinite 'Text' of repeated applications

998+-- of @f@ to @x@:

999+--

1000+-- > iterate f x == [x, f x, f (f x), ...]

1001+iterate :: (Char -> Char) -> Char -> Text

1002+iterate f c = let t c' = Chunk (T.singleton c') (t (f c'))

1003+ in t c

8851004

8861005-- | /O(n)/ 'replicateChar' @n@ @c@ is a 'Text' of length @n@ with @c@ the

8871006-- value of every element. Subject to fusion.

@@ -936,6 +1055,22 @@ take i t0 = take' i t0

9361055 unstream (S.take n (stream t)) = take n t

9371056 #-}

9381057

1058+-- | /O(n)/ 'takeEnd' @n@ @t@ returns the suffix remaining after

1059+-- taking @n@ characters from the end of @t@.

1060+--

1061+-- Examples:

1062+--

1063+-- > takeEnd 3 "foobar" == "bar"

1064+takeEnd :: Int64 -> Text -> Text

1065+takeEnd n t0

1066+ | n <= 0 = empty

1067+ | otherwise = takeChunk n empty . L.reverse . toChunks $ t0

1068+ where takeChunk _ acc [] = acc

1069+ takeChunk i acc (t:ts)

1070+ | i <= l = chunk (T.takeEnd (fromIntegral i) t) acc

1071+ | otherwise = takeChunk (i-l) (Chunk t acc) ts

1072+ where l = fromIntegral (T.length t)

1073+

9391074-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the

9401075-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@

9411076-- is greater than the length of the 'Text'. Subject to fusion.

@@ -958,6 +1093,23 @@ drop i t0

9581093 unstream (S.drop n (stream t)) = drop n t

9591094 #-}

9601095

1096+-- | /O(n)/ 'dropEnd' @n@ @t@ returns the prefix remaining after

1097+-- dropping @n@ characters from the end of @t@.

1098+--

1099+-- Examples:

1100+--

1101+-- > dropEnd 3 "foobar" == "foo"

1102+dropEnd :: Int64 -> Text -> Text

1103+dropEnd n t0

1104+ | n <= 0 = t0

1105+ | otherwise = dropChunk n . L.reverse . toChunks $ t0

1106+ where dropChunk _ [] = empty

1107+ dropChunk m (t:ts)

1108+ | m >= l = dropChunk (m-l) ts

1109+ | otherwise = fromChunks . L.reverse $

1110+ T.dropEnd (fromIntegral m) t : ts

1111+ where l = fromIntegral (T.length t)

1112+

9611113-- | /O(n)/ 'dropWords' @n@ returns the suffix with @n@ 'Word16'

9621114-- values dropped, or the empty 'Text' if @n@ is greater than the

9631115-- number of 'Word16' values present.

@@ -968,7 +1120,7 @@ dropWords i t0

9681120 where drop' 0 ts = ts

9691121 drop' _ Empty = Empty

9701122 drop' n (Chunk (T.Text arr off len) ts)

971− | n < len' = chunk (textP arr (off+n') (len-n')) ts

1123+ | n < len' = chunk (text arr (off+n') (len-n')) ts

9721124 | otherwise = drop' (n - len') ts

9731125 where len' = fromIntegral len

9741126 n' = fromIntegral n

@@ -992,6 +1144,20 @@ takeWhile p t0 = takeWhile' t0

9921144"LAZY TEXT takeWhile -> unfused" [1] forall p t.

9931145 unstream (S.takeWhile p (stream t)) = takeWhile p t

9941146 #-}

1147+-- | /O(n)/ 'takeWhileEnd', applied to a predicate @p@ and a 'Text',

1148+-- returns the longest suffix (possibly empty) of elements that

1149+-- satisfy @p@.

1150+-- Examples:

1151+--

1152+-- > takeWhileEnd (=='o') "foo" == "oo"

1153+takeWhileEnd :: (Char -> Bool) -> Text -> Text

1154+takeWhileEnd p = takeChunk empty . L.reverse . toChunks

1155+ where takeChunk acc [] = acc

1156+ takeChunk acc (t:ts) = if T.length t' < T.length t

1157+ then (Chunk t' acc)

1158+ else takeChunk (Chunk t' acc) ts

1159+ where t' = T.takeWhileEnd p t

1160+{-# INLINE takeWhileEnd #-}

9951161

9961162-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

9971163-- 'takeWhile' @p@ @t@. Subject to fusion.

@@ -1010,9 +1176,11 @@ dropWhile p t0 = dropWhile' t0

10101176"LAZY TEXT dropWhile -> unfused" [1] forall p t.

10111177 unstream (S.dropWhile p (stream t)) = dropWhile p t

10121178 #-}

1179+

10131180-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

1014−-- dropping characters that fail the predicate @p@ from the end of

1181+-- dropping characters that satisfy the predicate @p@ from the end of

10151182-- @t@.

1183+--

10161184-- Examples:

10171185--

10181186-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1029,7 +1197,7 @@ dropWhileEnd p = go

10291197{-# INLINE dropWhileEnd #-}

10301198

10311199-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

1032−-- dropping characters that fail the predicate @p@ from both the

1200+-- dropping characters that satisfy the predicate @p@ from both the

10331201-- beginning and end of @t@. Subject to fusion.

10341202dropAround :: (Char -> Bool) -> Text -> Text

10351203dropAround p = dropWhile p . dropWhileEnd p

@@ -1079,8 +1247,8 @@ splitAtWord _ Empty = empty :*: empty

10791247splitAtWord x (Chunk c@(T.Text arr off len) cs)

10801248 | y >= len = let h :*: t = splitAtWord (x-fromIntegral len) cs

10811249 in Chunk c h :*: t

1082− | otherwise = chunk (textP arr off y) empty :*:

1083− chunk (textP arr (off+y) (len-y)) cs

1250+ | otherwise = chunk (text arr off y) empty :*:

1251+ chunk (text arr (off+y) (len-y)) cs

10841252 where y = fromIntegral x

10851253

10861254-- | /O(n+m)/ Find the first instance of @needle@ (which must be

@@ -1224,9 +1392,10 @@ tails ts@(Chunk t ts')

12241392-- copies to create substrings; they just construct new 'Text's that

12251393-- are slices of the original.

12261394

1227−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first

1228−-- 'Text' argument, consuming the delimiter. An empty delimiter is

1229−-- invalid, and will cause an error to be raised.

1395+-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'

1396+-- argument (which cannot be an empty string), consuming the

1397+-- delimiter. An empty delimiter is invalid, and will cause an error

1398+-- to be raised.

12301399--

12311400-- Examples:

12321401--

@@ -1239,13 +1408,18 @@ tails ts@(Chunk t ts')

12391408-- > intercalate s . splitOn s == id

12401409-- > splitOn (singleton c) == split (==c)

12411410--

1411+-- (Note: the string @s@ to split on above cannot be empty.)

1412+--

12421413-- This function is strict in its first argument, and lazy in its

12431414-- second.

12441415--

12451416-- In (unlikely) bad cases, this function's time complexity degrades

12461417-- towards /O(n*m)/.

1247−splitOn :: Text -- ^ Text to split on

1248− -> Text -- ^ Input text

1418+splitOn :: Text

1419+ -- ^ String to split on. If this string is empty, an error

1420+ -- will occur.

1421+ -> Text

1422+ -- ^ Input text.

12491423 -> [Text]

12501424splitOn pat src

12511425 | null pat = emptyError "splitOn"

Data/Text/Lazy/Builder.hs+3−3

@@ -8,7 +8,7 @@

88-- Module : Data.Text.Lazy.Builder

11−-- License : BSD3-style (see LICENSE)

11+-- License : BSD-style (see LICENSE)

1212--

1313-- Maintainer : Johan Tibell <johan.tibell@gmail.com>

1414-- Stability : experimental

@@ -34,7 +34,7 @@

3434--

3535-- > singleton 'a' <> singleton 'b' <> singleton 'c'

3636--

37−-- since the '<>' from recent versions of 'Data.Monoid' associates

37+-- since the '<>' from recent versions of 'Data.Monoid' associates

3838-- to the right.

3939

4040-----------------------------------------------------------------------------

@@ -55,4 +55,4 @@ module Data.Text.Lazy.Builder

5555 , flush

5656 ) where

5757

58−import Data.Text.Lazy.Builder.Internal

58+import Data.Text.Internal.Builder

Data/Text/Lazy/Builder/Int.hs+25−18

@@ -1,4 +1,5 @@

1−{-# LANGUAGE BangPatterns, CPP, MagicHash, RankNTypes, UnboxedTuples #-}

1+{-# LANGUAGE BangPatterns, CPP, MagicHash, RankNTypes, ScopedTypeVariables,

2+ UnboxedTuples #-}

23#if __GLASGOW_HASKELL__ >= 702

34{-# LANGUAGE Trustworthy #-}

45#endif

@@ -6,7 +7,7 @@

67-- Module: Data.Text.Lazy.Builder.Int

9−-- License: BSD3

10+-- License: BSD-style

1011-- Maintainer: Bryan O'Sullivan <bos@serpentine.com>

1112-- Stability: experimental

1213-- Portability: portable

@@ -22,9 +23,9 @@ module Data.Text.Lazy.Builder.Int

2223import Data.Int (Int8, Int16, Int32, Int64)

2324import Data.Monoid (mempty)

2425import qualified Data.ByteString.Unsafe as B

25−import Data.Text.Lazy.Builder.Functions ((<>), i2d)

26−import Data.Text.Lazy.Builder.Internal

27−import Data.Text.Lazy.Builder.Int.Digits (digits)

26+import Data.Text.Internal.Builder.Functions ((<>), i2d)

27+import Data.Text.Internal.Builder

28+import Data.Text.Internal.Builder.Int.Digits (digits)

2829import Data.Text.Array

2930import Data.Word (Word, Word8, Word16, Word32, Word64)

3031import GHC.Base (quotInt, remInt)

@@ -34,7 +35,7 @@ import Control.Monad.ST

3435

3536#ifdef __GLASGOW_HASKELL__

3637# if defined(INTEGER_GMP)

37−import GHC.Integer.GMP.Internals

38+import GHC.Integer.GMP.Internals (Integer(S#))

3839# elif defined(INTEGER_SIMPLE)

3940import GHC.Integer

4041# else

@@ -49,18 +50,19 @@ import GHC.Integer

4950#endif

5051

5152decimal :: Integral a => a -> Builder

52−{-# SPECIALIZE decimal :: Int8 -> Builder #-}

53+{-# RULES "decimal/Int8" decimal = boundedDecimal :: Int8 -> Builder #-}

5354{-# RULES "decimal/Int" decimal = boundedDecimal :: Int -> Builder #-}

5455{-# RULES "decimal/Int16" decimal = boundedDecimal :: Int16 -> Builder #-}

5556{-# RULES "decimal/Int32" decimal = boundedDecimal :: Int32 -> Builder #-}

5657{-# RULES "decimal/Int64" decimal = boundedDecimal :: Int64 -> Builder #-}

57−{-# RULES "decimal/Word" decimal = positive :: Word -> Builder #-}

58+{-# RULES "decimal/Word" decimal = positive :: Data.Word.Word -> Builder #-}

5859{-# RULES "decimal/Word8" decimal = positive :: Word8 -> Builder #-}

5960{-# RULES "decimal/Word16" decimal = positive :: Word16 -> Builder #-}

6061{-# RULES "decimal/Word32" decimal = positive :: Word32 -> Builder #-}

6162{-# RULES "decimal/Word64" decimal = positive :: Word64 -> Builder #-}

6263{-# RULES "decimal/Integer" decimal = integer 10 :: Integer -> Builder #-}

6364decimal i = decimal' (<= -128) i

65+{-# NOINLINE decimal #-}

6466

6567boundedDecimal :: (Integral a, Bounded a) => a -> Builder

6668{-# SPECIALIZE boundedDecimal :: Int -> Builder #-}

@@ -132,8 +134,15 @@ i2w v = zero + fromIntegral v

132134

133135countDigits :: (Integral a) => a -> Int

134136{-# INLINE countDigits #-}

135−countDigits v0 = go 1 (fromIntegral v0 :: Word64)

136− where go !k v

137+countDigits v0

138+ | fromIntegral v64 == v0 = go 1 v64

139+ | otherwise = goBig 1 (fromIntegral v0)

140+ where v64 = fromIntegral v0

141+ goBig !k (v :: Integer)

142+ | v > big = goBig (k + 19) (v `quot` big)

143+ | otherwise = go k (fromIntegral v)

144+ big = 10000000000000000000

145+ go !k (v :: Word64)

137146 | v < 10 = k

138147 | v < 100 = k + 1

139148 | v < 1000 = k + 2

@@ -169,6 +178,7 @@ hexadecimal i

169178 where

170179 go n | n < 16 = hexDigit n

171180 | otherwise = go (n `quot` 16) <> hexDigit (n `rem` 16)

181+{-# NOINLINE[0] hexadecimal #-}

172182

173183hexInteger :: Integer -> Builder

174184hexInteger i

@@ -184,19 +194,12 @@ hexDigit n

184194 | otherwise = singleton $! toEnum (fromIntegral n + 87)

185195{-# INLINE hexDigit #-}

186196

187−int :: Int -> Builder

188−int = decimal

189−{-# INLINE int #-}

190−

191197data T = T !Integer !Int

192198

193199integer :: Int -> Integer -> Builder

194200#ifdef INTEGER_GMP

195201integer 10 (S# i#) = decimal (I# i#)

196202integer 16 (S# i#) = hexadecimal (I# i#)

197−#else

198−integer 10 i = decimal i

199−integer 16 i = hexadecimal i

200203#endif

201204integer base i

202205 | i < 0 = singleton '-' <> go (-i)

@@ -244,7 +247,11 @@ integer base i

244247 PAIR(x,y) -> pblock q <> pblock r <> putB ns

245248 where q = fromInteger x

246249 r = fromInteger y

247− putB _ = mempty

250+ putB _ = Data.Monoid.mempty

251+

252+ int :: Int -> Builder

253+ int x | base == 10 = decimal x

254+ | otherwise = hexadecimal x

248255

249256 pblock = loop maxDigits

250257 where

Data/Text/Lazy/Builder/RealFloat.hs+2−2

@@ -19,9 +19,9 @@ module Data.Text.Lazy.Builder.RealFloat

1919

2020import Data.Array.Base (unsafeAt)

2121import Data.Array.IArray

22−import Data.Text.Lazy.Builder.Functions ((<>), i2d)

22+import Data.Text.Internal.Builder.Functions ((<>), i2d)

2323import Data.Text.Lazy.Builder.Int (decimal)

24−import Data.Text.Lazy.Builder.RealFloat.Functions (roundTo)

24+import Data.Text.Internal.Builder.RealFloat.Functions (roundTo)

2525import Data.Text.Lazy.Builder

2626import qualified Data.Text as T

2727

Data/Text/Lazy/Builder/RealFloat/Functions.hsdeleted−25

@@ -1,25 +0,0 @@

1−-- |

2−-- Module: Data.Text.Lazy.Builder.RealFloat.Functions

3−-- Copyright: (c) The University of Glasgow 1994-2002

4−-- License: see libraries/base/LICENSE

5−

6−module Data.Text.Lazy.Builder.RealFloat.Functions

7− (

8− roundTo

9− ) where

10−

11−roundTo :: Int -> [Int] -> (Int,[Int])

12−roundTo d is =

13− case f d is of

14− x@(0,_) -> x

15− (1,xs) -> (1, 1:xs)

16− _ -> error "roundTo: bad Value"

17− where

18− f n [] = (0, replicate n 0)

19− f 0 (x:_) = (if x >= 5 then 1 else 0, [])

20− f n (i:xs)

21− | i' == 10 = (1,0:ds)

22− | otherwise = (0,i':ds)

23− where

24− (c,ds) = f (n-1) xs

25− i' = c + i

Data/Text/Lazy/Encoding.hs+35−11

@@ -7,8 +7,7 @@

88--

99-- License : BSD-style

10−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

11−-- duncan@haskell.org

10+-- Maintainer : bos@serpentine.com

1211-- Stability : experimental

1312-- Portability : portable

1413--

@@ -46,19 +45,28 @@ module Data.Text.Lazy.Encoding

4645 , encodeUtf16BE

4746 , encodeUtf32LE

4847 , encodeUtf32BE

48+

49+ -- * Encoding Text using ByteString Builders

50+ , encodeUtf8Builder

51+ , encodeUtf8BuilderEscaped

4952 ) where

5053

5154import Control.Exception (evaluate, try)

55+import Data.Monoid (Monoid(..))

5256import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

53−import Data.Text.Lazy.Internal (Text(..), chunk, empty, foldrChunks)

57+import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldrChunks)

58+import Data.Word (Word8)

5459import qualified Data.ByteString as S

60+import qualified Data.ByteString.Builder as B

61+import qualified Data.ByteString.Builder.Extra as B (safeStrategy, toLazyByteStringWith)

62+import qualified Data.ByteString.Builder.Prim as BP

5563import qualified Data.ByteString.Lazy as B

5664import qualified Data.ByteString.Lazy.Internal as B

5765import qualified Data.ByteString.Unsafe as B

66+import qualified Data.Text as T

5867import qualified Data.Text.Encoding as TE

59−import qualified Data.Text.Lazy as L

60−import qualified Data.Text.Lazy.Encoding.Fusion as E

61−import qualified Data.Text.Lazy.Fusion as F

68+import qualified Data.Text.Internal.Lazy.Encoding.Fusion as E

69+import qualified Data.Text.Internal.Lazy.Fusion as F

6270import Data.Text.Unsafe (unsafeDupablePerformIO)

6371

6472-- $strict

@@ -76,8 +84,6 @@ import Data.Text.Unsafe (unsafeDupablePerformIO)

7684

7785-- | /Deprecated/. Decode a 'ByteString' containing 7-bit ASCII

7886-- encoded text.

79−--

80−-- This function is deprecated. Use 'decodeLatin1' instead.

8187decodeASCII :: B.ByteString -> Text

8288decodeASCII = decodeUtf8

8389{-# DEPRECATED decodeASCII "Use decodeUtf8 instead" #-}

@@ -99,7 +105,7 @@ decodeUtf8With onErr (B.Chunk b0 bs0) =

99105 | S.null l = empty

100106 | otherwise = case onErr desc (Just (B.unsafeHead l)) of

101107 Nothing -> empty

102− Just c -> L.singleton c

108+ Just c -> Chunk (T.singleton c) Empty

103109 desc = "Data.Text.Lazy.Encoding.decodeUtf8With: Invalid UTF-8 stream"

104110decodeUtf8With _ _ = empty

105111

@@ -136,8 +142,26 @@ decodeUtf8' bs = unsafeDupablePerformIO $ do

136142{-# INLINE decodeUtf8' #-}

137143

138144encodeUtf8 :: Text -> B.ByteString

139−encodeUtf8 (Chunk c cs) = B.Chunk (TE.encodeUtf8 c) (encodeUtf8 cs)

140−encodeUtf8 Empty = B.Empty

145+encodeUtf8 Empty = B.empty

146+encodeUtf8 lt@(Chunk t _) =

147+ B.toLazyByteStringWith strategy B.empty $ encodeUtf8Builder lt

148+ where

149+ -- To improve our small string performance, we use a strategy that

150+ -- allocates a buffer that is guaranteed to be large enough for the

151+ -- encoding of the first chunk, but not larger than the default

152+ -- B.smallChunkSize. We clamp the firstChunkSize to ensure that we don't

153+ -- generate too large buffers which hamper streaming.

154+ firstChunkSize = min B.smallChunkSize (4 * (T.length t + 1))

155+ strategy = B.safeStrategy firstChunkSize B.defaultChunkSize

156+

157+encodeUtf8Builder :: Text -> B.Builder

158+encodeUtf8Builder =

159+ foldrChunks (\c b -> TE.encodeUtf8Builder c `mappend` b) Data.Monoid.mempty

160+

161+{-# INLINE encodeUtf8BuilderEscaped #-}

162+encodeUtf8BuilderEscaped :: BP.BoundedPrim Word8 -> Text -> B.Builder

163+encodeUtf8BuilderEscaped prim =

164+ foldrChunks (\c b -> TE.encodeUtf8BuilderEscaped prim c `mappend` b) mempty

141165

142166-- | Decode text from little endian UTF-16 encoding.

143167decodeUtf16LEWith :: OnDecodeError -> B.ByteString -> Text

Data/Text/Lazy/IO.hs+2−2

@@ -50,8 +50,8 @@ import qualified Data.Text.Lazy as L

5050import qualified Control.Exception as E

5151import Control.Monad (when)

5252import Data.IORef (readIORef)

53−import Data.Text.IO.Internal (hGetLineWith, readChunk)

54−import Data.Text.Lazy.Internal (chunk, empty)

53+import Data.Text.Internal.IO (hGetLineWith, readChunk)

54+import Data.Text.Internal.Lazy (chunk, empty)

5555import GHC.IO.Buffer (isEmptyBuffer)

5656import GHC.IO.Exception (IOException(..), IOErrorType(..), ioException)

5757import GHC.IO.Handle.Internals (augmentIOError, hClose_help,

Data/Text/Lazy/Internal.hs+7−105

@@ -1,118 +1,20 @@

11{-# LANGUAGE BangPatterns, DeriveDataTypeable #-}

22-- |

33-- Module : Data.Text.Lazy.Internal

55--

66-- License : BSD-style

7−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

8−-- duncan@haskell.org

7+-- Maintainer : bos@serpentine.com

98-- Stability : experimental

109-- Portability : GHC

1110--

12−-- A module containing private 'Text' internals. This exposes the

13−-- 'Text' representation and low level construction functions.

14−-- Modules which extend the 'Text' system may need to use this module.

15−--

16−-- You should not use this module unless you are determined to monkey

17−-- with the internals, as the functions here do just about nothing to

18−-- preserve data invariants. You have been warned!

11+-- This module has been renamed to 'Data.Text.Internal.Lazy'. This

12+-- name for the module will be removed in the next major release.

1913

2014module Data.Text.Lazy.Internal

15+ {-# DEPRECATED "Use Data.Text.Internal.Lazy instead" #-}

2116 (

22− Text(..)

23− , chunk

24− , empty

25− , foldrChunks

26− , foldlChunks

27− -- * Data type invariant and abstraction functions

28−

29− -- $invariant

30− , strictInvariant

31− , lazyInvariant

32− , showStructure

33−

34− -- * Chunk allocation sizes

35− , defaultChunkSize

36− , smallChunkSize

37− , chunkOverhead

17+ module Data.Text.Internal.Lazy

3818 ) where

3919

40−import Data.Text ()

41−import Data.Text.UnsafeShift (shiftL)

42−import Data.Typeable (Typeable)

43−import Foreign.Storable (sizeOf)

44−import qualified Data.Text.Internal as T

45−

46−data Text = Empty

47− | Chunk {-# UNPACK #-} !T.Text Text

48− deriving (Typeable)

49−

50−-- $invariant

51−--

52−-- The data type invariant for lazy 'Text': Every 'Text' is either 'Empty' or

53−-- consists of non-null 'T.Text's. All functions must preserve this,

54−-- and the QC properties must check this.

55−

56−-- | Check the invariant strictly.

57−strictInvariant :: Text -> Bool

58−strictInvariant Empty = True

59−strictInvariant x@(Chunk (T.Text _ _ len) cs)

60− | len > 0 = strictInvariant cs

61− | otherwise = error $ "Data.Text.Lazy: invariant violation: "

62− ++ showStructure x

63−

64−-- | Check the invariant lazily.

65−lazyInvariant :: Text -> Text

66−lazyInvariant Empty = Empty

67−lazyInvariant x@(Chunk c@(T.Text _ _ len) cs)

68− | len > 0 = Chunk c (lazyInvariant cs)

69− | otherwise = error $ "Data.Text.Lazy: invariant violation: "

70− ++ showStructure x

71−

72−-- | Display the internal structure of a lazy 'Text'.

73−showStructure :: Text -> String

74−showStructure Empty = "Empty"

75−showStructure (Chunk t Empty) = "Chunk " ++ show t ++ " Empty"

76−showStructure (Chunk t ts) =

77− "Chunk " ++ show t ++ " (" ++ showStructure ts ++ ")"

78−

79−-- | Smart constructor for 'Chunk'. Guarantees the data type invariant.

80−chunk :: T.Text -> Text -> Text

81−{-# INLINE chunk #-}

82−chunk t@(T.Text _ _ len) ts | len == 0 = ts

83− | otherwise = Chunk t ts

84−

85−-- | Smart constructor for 'Empty'.

86−empty :: Text

87−{-# INLINE [0] empty #-}

88−empty = Empty

89−

90−-- | Consume the chunks of a lazy 'Text' with a natural right fold.

91−foldrChunks :: (T.Text -> a -> a) -> a -> Text -> a

92−foldrChunks f z = go

93− where go Empty = z

94− go (Chunk c cs) = f c (go cs)

95−{-# INLINE foldrChunks #-}

96−

97−-- | Consume the chunks of a lazy 'Text' with a strict, tail-recursive,

98−-- accumulating left fold.

99−foldlChunks :: (a -> T.Text -> a) -> a -> Text -> a

100−foldlChunks f z = go z

101− where go !a Empty = a

102− go !a (Chunk c cs) = go (f a c) cs

103−{-# INLINE foldlChunks #-}

104−

105−-- | Currently set to 16 KiB, less the memory management overhead.

106−defaultChunkSize :: Int

107−defaultChunkSize = 16384 - chunkOverhead

108−{-# INLINE defaultChunkSize #-}

109−

110−-- | Currently set to 128 bytes, less the memory management overhead.

111−smallChunkSize :: Int

112−smallChunkSize = 128 - chunkOverhead

113−{-# INLINE smallChunkSize #-}

114−

115−-- | The memory management overhead. Currently this is tuned for GHC only.

116−chunkOverhead :: Int

117−chunkOverhead = sizeOf (undefined :: Int) `shiftL` 1

118−{-# INLINE chunkOverhead #-}

20+import Data.Text.Internal.Lazy

Data/Text/Lazy/Read.hs+8−33

@@ -1,5 +1,7 @@

11{-# LANGUAGE OverloadedStrings, CPP #-}

2−#if __GLASGOW_HASKELL__ >= 702

2+#if __GLASGOW_HASKELL__ >= 704

3+{-# LANGUAGE Safe #-}

4+#elif __GLASGOW_HASKELL__ >= 702

35{-# LANGUAGE Trustworthy #-}

46#endif

@@ -24,15 +26,17 @@ module Data.Text.Lazy.Read

2426 ) where

2527

2628import Control.Monad (liftM)

27−import Data.Char (isDigit, isHexDigit, ord)

29+import Data.Char (isDigit, isHexDigit)

2830import Data.Int (Int8, Int16, Int32, Int64)

2931import Data.Ratio ((%))

32+import Data.Text.Internal.Read

3033import Data.Text.Lazy as T

3134import Data.Word (Word, Word8, Word16, Word32, Word64)

3235

3336-- | Read some text. If the read succeeds, return its value and the

3437-- remaining text, otherwise an error message.

35−type Reader a = Text -> Either String (a,Text)

38+type Reader a = IReader Text a

39+type Parser = IParser Text

3640

3741-- | Read a decimal integer. The input must begin with at least one

3842-- decimal digit, and is consumed until a non-digit or end of string

@@ -52,7 +56,7 @@ decimal :: Integral a => Reader a

5256{-# SPECIALIZE decimal :: Reader Int32 #-}

5357{-# SPECIALIZE decimal :: Reader Int64 #-}

5458{-# SPECIALIZE decimal :: Reader Integer #-}

55−{-# SPECIALIZE decimal :: Reader Word #-}

59+{-# SPECIALIZE decimal :: Reader Data.Word.Word #-}

5660{-# SPECIALIZE decimal :: Reader Word8 #-}

5761{-# SPECIALIZE decimal :: Reader Word16 #-}

5862{-# SPECIALIZE decimal :: Reader Word32 #-}

@@ -101,15 +105,6 @@ hex txt

101105 where (h,t) = T.span isHexDigit txt

102106 go n d = (n * 16 + fromIntegral (hexDigitToInt d))

103107

104−hexDigitToInt :: Char -> Int

105−hexDigitToInt c

106− | c >= '0' && c <= '9' = ord c - ord '0'

107− | c >= 'a' && c <= 'f' = ord c - (ord 'a' - 10)

108− | otherwise = ord c - (ord 'A' - 10)

109−

110−digitToInt :: Char -> Int

111−digitToInt c = ord c - ord '0'

112−

113108-- | Read an optional leading sign character (@\'-\'@ or @\'+\'@) and

114109-- apply it to the result of applying the given reader.

115110signed :: Num a => Reader a -> Reader a

@@ -169,31 +164,11 @@ signa p = do

169164 sign <- perhaps '+' $ char (\c -> c == '-' || c == '+')

170165 if sign == '+' then p else negate `liftM` p

171166

172−newtype Parser a = P {

173− runP :: Reader a

174− }

175−

176−instance Monad Parser where

177− return a = P $ \t -> Right (a,t)

178− {-# INLINE return #-}

179− m >>= k = P $ \t -> case runP m t of

180− Left err -> Left err

181− Right (a,t') -> runP (k a) t'

182− {-# INLINE (>>=) #-}

183− fail msg = P $ \_ -> Left msg

184−

185−perhaps :: a -> Parser a -> Parser a

186−perhaps def m = P $ \t -> case runP m t of

187− Left _ -> Right (def,t)

188− r@(Right _) -> r

189−

190167char :: (Char -> Bool) -> Parser Char

191168char p = P $ \t -> case T.uncons t of

192169 Just (c,t') | p c -> Right (c,t')

193170 _ -> Left "character does not match"

194171

195−data T = T !Integer !Int

196−

197172floaty :: Fractional a => (Integer -> Integer -> Integer -> a) -> Reader a

198173{-# INLINE floaty #-}

199174floaty f = runP $ do

Data/Text/Read.hs+6−33

@@ -24,16 +24,18 @@ module Data.Text.Read

2424 ) where

2525

2626import Control.Monad (liftM)

27−import Data.Char (isDigit, isHexDigit, ord)

27+import Data.Char (isDigit, isHexDigit)

2828import Data.Int (Int8, Int16, Int32, Int64)

2929import Data.Ratio ((%))

3030import Data.Text as T

31−import Data.Text.Private (span_)

31+import Data.Text.Internal.Private (span_)

32+import Data.Text.Internal.Read

3233import Data.Word (Word, Word8, Word16, Word32, Word64)

3334

3435-- | Read some text. If the read succeeds, return its value and the

3536-- remaining text, otherwise an error message.

36−type Reader a = Text -> Either String (a,Text)

37+type Reader a = IReader Text a

38+type Parser a = IParser Text a

3739

3840-- | Read a decimal integer. The input must begin with at least one

3941-- decimal digit, and is consumed until a non-digit or end of string

@@ -53,7 +55,7 @@ decimal :: Integral a => Reader a

5355{-# SPECIALIZE decimal :: Reader Int32 #-}

5456{-# SPECIALIZE decimal :: Reader Int64 #-}

5557{-# SPECIALIZE decimal :: Reader Integer #-}

56−{-# SPECIALIZE decimal :: Reader Word #-}

58+{-# SPECIALIZE decimal :: Reader Data.Word.Word #-}

5759{-# SPECIALIZE decimal :: Reader Word8 #-}

5860{-# SPECIALIZE decimal :: Reader Word16 #-}

5961{-# SPECIALIZE decimal :: Reader Word32 #-}

@@ -111,15 +113,6 @@ hex txt

111113 where (# h,t #) = span_ isHexDigit txt

112114 go n d = (n * 16 + fromIntegral (hexDigitToInt d))

113115

114−hexDigitToInt :: Char -> Int

115−hexDigitToInt c

116− | c >= '0' && c <= '9' = ord c - ord '0'

117− | c >= 'a' && c <= 'f' = ord c - (ord 'a' - 10)

118− | otherwise = ord c - (ord 'A' - 10)

119−

120−digitToInt :: Char -> Int

121−digitToInt c = ord c - ord '0'

122−

123116-- | Read an optional leading sign character (@\'-\'@ or @\'+\'@) and

124117-- apply it to the result of applying the given reader.

125118signed :: Num a => Reader a -> Reader a

@@ -179,31 +172,11 @@ signa p = do

179172 sign <- perhaps '+' $ char (\c -> c == '-' || c == '+')

180173 if sign == '+' then p else negate `liftM` p

181174

182−newtype Parser a = P {

183− runP :: Reader a

184− }

185−

186−instance Monad Parser where

187− return a = P $ \t -> Right (a,t)

188− {-# INLINE return #-}

189− m >>= k = P $ \t -> case runP m t of

190− Left err -> Left err

191− Right (a,t') -> runP (k a) t'

192− {-# INLINE (>>=) #-}

193− fail msg = P $ \_ -> Left msg

194−

195−perhaps :: a -> Parser a -> Parser a

196−perhaps def m = P $ \t -> case runP m t of

197− Left _ -> Right (def,t)

198− r@(Right _) -> r

199−

200175char :: (Char -> Bool) -> Parser Char

201176char p = P $ \t -> case T.uncons t of

202177 Just (c,t') | p c -> Right (c,t')

203178 _ -> Left "character does not match"

204179

205−data T = T !Integer !Int

206−

207180floaty :: Fractional a => (Integer -> Integer -> Integer -> a) -> Reader a

208181{-# INLINE floaty #-}

209182floaty f = runP $ do

Data/Text/Show.hsnew file+89

@@ -0,0 +1,89 @@

1+{-# LANGUAGE CPP, MagicHash #-}

2+{-# OPTIONS_GHC -fno-warn-orphans #-}

3+#if __GLASGOW_HASKELL__ >= 702

4+{-# LANGUAGE Trustworthy #-}

5+#endif

7+-- |

8+-- Module : Data.Text.Show

10+--

11+-- License : BSD-style

12+-- Maintainer : bos@serpentine.com

13+-- Stability : experimental

14+-- Portability : GHC

15+

16+module Data.Text.Show

17+ (

18+ singleton

19+ , unpack

20+ , unpackCString#

21+ ) where

22+

23+import Control.Monad.ST (ST)

24+import Data.Text.Internal (Text(..), empty_, safe)

25+import Data.Text.Internal.Fusion (stream, unstream)

26+import Data.Text.Internal.Unsafe.Char (unsafeWrite)

27+import GHC.Prim (Addr#)

28+import qualified Data.Text.Array as A

29+import qualified Data.Text.Internal.Fusion.Common as S

30+

31+#if __GLASGOW_HASKELL__ >= 702

32+import qualified GHC.CString as GHC

33+#else

34+import qualified GHC.Base as GHC

35+#endif

36+

37+instance Show Text where

38+ showsPrec p ps r = showsPrec p (unpack ps) r

39+

40+-- | /O(n)/ Convert a 'Text' into a 'String'. Subject to fusion.

41+unpack :: Text -> String

42+unpack = S.unstreamList . stream

43+{-# INLINE [1] unpack #-}

44+

45+-- | /O(n)/ Convert a literal string into a 'Text'. Subject to

46+-- fusion.

47+--

48+-- This is exposed solely for people writing GHC rewrite rules.

49+unpackCString# :: Addr# -> Text

50+unpackCString# addr# = unstream (S.streamCString# addr#)

51+{-# NOINLINE unpackCString# #-}

52+

53+{-# RULES "TEXT literal" [1] forall a.

54+ unstream (S.map safe (S.streamList (GHC.unpackCString# a)))

55+ = unpackCString# a #-}

56+

57+{-# RULES "TEXT literal UTF8" [1] forall a.

58+ unstream (S.map safe (S.streamList (GHC.unpackCStringUtf8# a)))

59+ = unpackCString# a #-}

60+

61+{-# RULES "TEXT empty literal" [1]

62+ unstream (S.map safe (S.streamList []))

63+ = empty_ #-}

64+

65+{-# RULES "TEXT singleton literal" [1] forall a.

66+ unstream (S.map safe (S.streamList [a]))

67+ = singleton_ a #-}

68+

69+-- | /O(1)/ Convert a character into a Text. Subject to fusion.

70+-- Performs replacement on invalid scalar values.

71+singleton :: Char -> Text

72+singleton = unstream . S.singleton . safe

73+{-# INLINE [1] singleton #-}

74+

75+{-# RULES "TEXT singleton" forall a.

76+ unstream (S.singleton (safe a))

77+ = singleton_ a #-}

78+

79+-- This is intended to reduce inlining bloat.

80+singleton_ :: Char -> Text

81+singleton_ c = Text (A.run x) 0 len

82+ where x :: ST s (A.MArray s)

83+ x = do arr <- A.new len

84+ _ <- unsafeWrite arr 0 d

85+ return arr

86+ len | d < '\x10000' = 1

87+ | otherwise = 2

88+ d = safe c

89+{-# NOINLINE singleton_ #-}

Data/Text/Unsafe.hs+15−5

@@ -3,8 +3,7 @@

33-- Module : Data.Text.Unsafe

55-- License : BSD-style

6−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

7−-- duncan@haskell.org

6+-- Maintainer : bos@serpentine.com

87-- Stability : experimental

98-- Portability : portable

109--

@@ -19,6 +18,7 @@ module Data.Text.Unsafe

1918 , iter

2019 , iter_

2120 , reverseIter

21+ , reverseIter_

2222 , unsafeHead

2323 , unsafeTail

2424 , lengthWord16

@@ -29,10 +29,10 @@ module Data.Text.Unsafe

2929#if defined(ASSERTS)

3030import Control.Exception (assert)

3131#endif

32−import Data.Text.Encoding.Utf16 (chr2)

32+import Data.Text.Internal.Encoding.Utf16 (chr2)

3333import Data.Text.Internal (Text(..))

34−import Data.Text.Unsafe.Base (inlineInterleaveST, inlinePerformIO)

35−import Data.Text.UnsafeChar (unsafeChr)

34+import Data.Text.Internal.Unsafe (inlineInterleaveST, inlinePerformIO)

35+import Data.Text.Internal.Unsafe.Char (unsafeChr)

3636import qualified Data.Text.Array as A

3737import GHC.IO (unsafeDupablePerformIO)

3838

@@ -95,6 +95,16 @@ reverseIter (Text arr off _len) i

9595 k = j - 1

9696{-# INLINE reverseIter #-}

9797

98+-- | /O(1)/ Iterate one step backwards through a UTF-16 array,

99+-- returning the delta to add (i.e. a negative number) to give the

100+-- next offset to iterate at.

101+reverseIter_ :: Text -> Int -> Int

102+reverseIter_ (Text arr off _len) i

103+ | m < 0xDC00 || m > 0xDFFF = -1

104+ | otherwise = -2

105+ where m = A.unsafeIndex arr (off+i)

106+{-# INLINE reverseIter_ #-}

107+

98108-- | /O(1)/ Return the length of a 'Text' in units of 'Word16'. This

99109-- is useful for sizing a target array appropriately before using

100110-- 'unsafeCopyToPtr'.

benchmarks/haskell/Benchmarks.hs+6−1

@@ -15,6 +15,7 @@ import qualified Benchmarks.EncodeUtf8 as EncodeUtf8

1515import qualified Benchmarks.Equality as Equality

1616import qualified Benchmarks.FileRead as FileRead

1717import qualified Benchmarks.FoldLines as FoldLines

18+import qualified Benchmarks.Mul as Mul

1819import qualified Benchmarks.Pure as Pure

1920import qualified Benchmarks.ReadNumbers as ReadNumbers

2021import qualified Benchmarks.Replace as Replace

@@ -49,7 +50,11 @@ benchmarks = do

4950 , Equality.benchmark (tf "japanese.txt")

5051 , FileRead.benchmark (tf "russian.txt")

5152 , FoldLines.benchmark (tf "russian.txt")

52− , Pure.benchmark "tiny "(tf "tiny.txt")

53+ , Mul.benchmark

54+ , Pure.benchmark "tiny" (tf "tiny.txt")

55+ , Pure.benchmark "ascii" (tf "ascii-small.txt")

56+ -- , Pure.benchmark "france" (tf "france.html")

57+ , Pure.benchmark "russian" (tf "russian-small.txt")

5358 , Pure.benchmark "japanese" (tf "japanese.txt")

5459 , ReadNumbers.benchmark (tf "numbers.txt")

5560 , Replace.benchmark (tf "russian.txt") "принимая" "своем"

benchmarks/haskell/Benchmarks/DecodeUtf8.hs+2−2

@@ -25,7 +25,7 @@ import Foreign.Ptr (Ptr, plusPtr)

2525import Foreign.ForeignPtr (withForeignPtr)

2626import Data.Word (Word8)

2727import qualified Criterion as C

28−import Criterion (Benchmark, bgroup, nf)

28+import Criterion (Benchmark, bgroup, nf, whnfIO)

2929import qualified Codec.Binary.UTF8.Generic as U8

3030import qualified Data.ByteString as B

3131import qualified Data.ByteString.Lazy as BL

@@ -48,7 +48,7 @@ benchmark kind fp = do

4848 return $ bgroup "DecodeUtf8"

4949 [ bench "Strict" $ nf T.decodeUtf8 bs

5050 , bench "Stream" $ nf decodeStream lbs

51− , bench "IConv" $ iconv bs

51+ , bench "IConv" $ whnfIO $ iconv bs

5252 , bench "StrictLength" $ nf (T.length . T.decodeUtf8) bs

5353 , bench "StrictInitLength" $ nf (T.length . T.init . T.decodeUtf8) bs

5454 , bench "Lazy" $ nf TL.decodeUtf8 lbs

benchmarks/haskell/Benchmarks/FileRead.hs+11−11

@@ -8,8 +8,8 @@ module Benchmarks.FileRead

88 ( benchmark

99 ) where

1010

11−import Control.Exception (evaluate)

12−import Criterion (Benchmark, bgroup, bench)

11+import Control.Applicative ((<$>))

12+import Criterion (Benchmark, bgroup, bench, whnfIO)

1313import qualified Data.ByteString as SB

1414import qualified Data.ByteString.Lazy as LB

1515import qualified Data.Text as T

@@ -21,13 +21,13 @@ import qualified Data.Text.Lazy.IO as LT

2121

2222benchmark :: FilePath -> IO Benchmark

2323benchmark p = return $ bgroup "FileRead"

24− [ bench "String" $ readFile p >>= evaluate . length

25− , bench "ByteString" $ SB.readFile p >>= evaluate . SB.length

26− , bench "LazyByteString" $ LB.readFile p >>= evaluate . LB.length

27− , bench "Text" $ T.readFile p >>= evaluate . T.length

28− , bench "LazyText" $ LT.readFile p >>= evaluate . LT.length

29− , bench "TextByteString" $

30− SB.readFile p >>= evaluate . T.length . T.decodeUtf8

31− , bench "LazyTextByteString" $

32− LB.readFile p >>= evaluate . LT.length . LT.decodeUtf8

24+ [ bench "String" $ whnfIO $ length <$> readFile p

25+ , bench "ByteString" $ whnfIO $ SB.length <$> SB.readFile p

26+ , bench "LazyByteString" $ whnfIO $ LB.length <$> LB.readFile p

27+ , bench "Text" $ whnfIO $ T.length <$> T.readFile p

28+ , bench "LazyText" $ whnfIO $ LT.length <$> LT.readFile p

29+ , bench "TextByteString" $ whnfIO $

30+ (T.length . T.decodeUtf8) <$> SB.readFile p

31+ , bench "LazyTextByteString" $ whnfIO $

32+ (LT.length . LT.decodeUtf8) <$> LB.readFile p

3333 ]

benchmarks/haskell/Benchmarks/FoldLines.hs+2−2

@@ -10,7 +10,7 @@ module Benchmarks.FoldLines

1010 ( benchmark

1111 ) where

1212

13−import Criterion (Benchmark, bgroup, bench)

13+import Criterion (Benchmark, bgroup, bench, whnfIO)

1414import System.IO

1515import qualified Data.ByteString as B

1616import qualified Data.Text as T

@@ -22,7 +22,7 @@ benchmark fp = return $ bgroup "ReadLines"

2222 , bench "ByteString" $ withHandle $ foldLinesB (\n _ -> n + 1) (0 :: Int)

2323 ]

2424 where

25− withHandle f = do

25+ withHandle f = whnfIO $ do

2626 h <- openFile fp ReadMode

2727 hSetBuffering h (BlockBuffering (Just 16384))

2828 x <- f h

benchmarks/haskell/Benchmarks/Mul.hsnew file+138

@@ -0,0 +1,138 @@

1+module Benchmarks.Mul (benchmark) where

3+import Control.Exception (evaluate)

4+import Criterion.Main

5+import Data.Int (Int32, Int64)

6+import Data.Text.Internal (mul32, mul64)

7+import qualified Data.Vector.Unboxed as U

9+oldMul :: Int64 -> Int64 -> Int64

10+oldMul m n

11+ | n == 0 = 0

12+ | m <= maxBound `quot` n = m * n

13+ | otherwise = error "overflow"

14+

15+benchmark :: IO Benchmark

16+benchmark = do

17+ _ <- evaluate testVector32

18+ _ <- evaluate testVector64

19+ return $ bgroup "Mul" [

20+ bench "oldMul" $ whnf (U.map (uncurry oldMul)) testVector64

21+ , bench "mul64" $ whnf (U.map (uncurry mul64)) testVector64

22+ , bench "*64" $ whnf (U.map (uncurry (*))) testVector64

23+ , bench "mul32" $ whnf (U.map (uncurry mul32)) testVector32

24+ , bench "*32" $ whnf (U.map (uncurry (*))) testVector32

25+ ]

26+

27+testVector64 :: U.Vector (Int64,Int64)

28+testVector64 = U.fromList [

29+ (0,1248868987182846646),(169004623633872,24458),(482549039517835,7614),

30+ (372,8157063115504364),(27,107095594861148252),(3,63249878517962420),

31+ (4363,255694473572912),(86678474,1732634806),(1572453024,1800489338),

32+ (9384523143,77053781),(49024709555,75095046),(7,43457620410239131),

33+ (8,8201563008844571),(387719037,1520696708),(189869238220197,1423),

34+ (46788016849611,23063),(503077742109974359,0),(104,1502010908706487),

35+ (30478140346,207525518),(80961140129236192,14),(4283,368012829143675),

36+ (1028719181728108146,6),(318904,5874863049591),(56724427166898,110794),

37+ (234539368,31369110449),(2,251729663598178612),(103291548194451219,5),

38+ (76013,5345328755566),(1769631,2980846129318),(40898,60598477385754),

39+ (0,98931348893227155),(573555872156917492,3),(318821187115,4476566),

40+ (11152874213584,243582),(40274276,16636653248),(127,4249988676030597),

41+ (103543712111871836,5),(71,16954462148248238),(3963027173504,216570),

42+ (13000,503523808916753),(17038308,20018685905),(0,510350226577891549),

43+ (175898,3875698895405),(425299191292676,5651),(17223451323664536,50),

44+ (61755131,14247665326),(0,1018195131697569303),(36433751497238985,20),

45+ (3473607861601050,1837),(1392342328,1733971838),(225770297367,3249655),

46+ (14,127545244155254102),(1751488975299136,2634),(3949208,504190668767),

47+ (153329,831454434345),(1066212122928663658,2),(351224,2663633539556),

48+ (344565,53388869217),(35825609350446863,54),(276011553660081475,10),

49+ (1969754174790470349,3),(35,68088438338633),(506710,3247689556438),

50+ (11099382291,327739909),(105787303549,32824363),(210366111,14759049409),

51+ (688893241579,3102676),(8490,70047474429581),(152085,29923000251880),

52+ (5046974599257095,400),(4183167795,263434071),(10089728,502781960687),

53+ (44831977765,4725378),(91,8978094664238578),(30990165721,44053350),

54+ (1772377,149651820860),(243420621763408572,4),(32,5790357453815138),

55+ (27980806337993771,5),(47696295759774,20848),(1745874142313778,1098),

56+ (46869334770121,1203),(886995283,1564424789),(40679396544,76002479),

57+ (1,672849481568486995),(337656187205,3157069),(816980552858963,6003),

58+ (2271434085804831543,1),(0,1934521023868747186),(6266220038281,15825),

59+ (4160,107115946987394),(524,246808621791561),(0,1952519482439636339),

60+ (128,2865935904539691),(1044,3211982069426297),(16000511542473,88922),

61+ (1253596745404082,2226),(27041,56836278958002),(23201,49247489754471),

62+ (175906590497,21252392),(185163584757182295,24),(34742225226802197,150),

63+ (2363228,250824838408),(216327527109550,45),(24,81574076994520675),

64+ (28559899906542,15356),(10890139774837133,511),(2293,707179303654492),

65+ (2749366833,40703233),(0,4498229704622845986),(439,4962056468281937),

66+ (662,1453820621089921),(16336770612459631,220),(24282989393,74239137),

67+ (2724564648490195,3),(743672760,124992589),(4528103,704330948891),

68+ (6050483122491561,250),(13322953,13594265152),(181794,22268101450214),

69+ (25957941712,75384092),(43352,7322262295009),(32838,52609059549923),

70+ (33003585202001564,2),(103019,68430142267402),(129918230800,8742978),

71+ (0,2114347379589080688),(2548,905723041545274),(222745067962838382,0),

72+ (1671683850790425181,1),(455,4836932776795684),(794227702827214,6620),

73+ (212534135175874,1365),(96432431858,29784975),(466626763743380,3484),

74+ (29793949,53041519613),(8359,309952753409844),(3908960585331901,26),

75+ (45185288970365760,114),(10131829775,68110174),(58039242399640479,83),

76+ (628092278238719399,6),(1,196469106875361889),(302336625,16347502444),

77+ (148,3748088684181047),(1,1649096568849015456),(1019866864,2349753026),

78+ (8211344830,569363306),(65647579546873,34753),(2340190,1692053129069),

79+ (64263301,30758930355),(48681618072372209,110),(7074794736,47640197),

80+ (249634721521,7991792),(1162917363807215,232),(7446433349,420634045),

81+ (63398619383,60709817),(51359004508011,14200),(131788797028647,7072),

82+ (52079887791430043,7),(7,136277667582599838),(28582879735696,50327),

83+ (1404582800566278,833),(469164435,15017166943),(99567079957578263,49),

84+ (1015285971,3625801566),(321504843,4104079293),(5196954,464515406632),

85+ (114246832260876,7468),(8149664437,487119673),(12265299,378168974869),

86+ (37711995764,30766513),(3971137243,710996152),(483120070302,603162),

87+ (103009942,61645547145),(8476344625340,6987),(547948761229739,1446),

88+ (42234,18624767306301),(13486714173011,58948),(4,198309153268019840),

89+ (9913176974,325539248),(28246225540203,116822),(2882463945582154,18),

90+ (959,25504987505398),(3,1504372236378217710),(13505229956793,374987),

91+ (751661959,457611342),(27375926,36219151769),(482168869,5301952074),

92+ (1,1577425863241520640),(714116235611821,1164),(904492524250310488,0),

93+ (5983514941763398,68),(10759472423,23540686),(72539568471529,34919),

94+ (4,176090672310337473),(938702842110356453,1),(673652445,3335287382),

95+ (3111998893666122,917),(1568013,3168419765469)]

96+

97+testVector32 :: U.Vector (Int32,Int32)

98+testVector32 = U.fromList [

99+ (39242,410),(0,100077553),(2206,9538),(509400240,1),(38048,6368),

100+ (1789,651480),(2399,157032),(701,170017),(5241456,14),(11212,70449),

101+ (1,227804876),(749687254,1),(74559,2954),(1158,147957),(410604456,1),

102+ (170851,1561),(92643422,1),(6192,180509),(7,24202210),(3440,241481),

103+ (5753677,5),(294327,1622),(252,4454673),(127684121,11),(28315800,30),

104+ (340370905,0),(1,667887987),(592782090,1),(49023,27641),(750,290387),

105+ (72886,3847),(0,301047933),(3050276,473),(1,788366142),(59457,15813),

106+ (637726933,1),(1135,344317),(853616,264),(696816,493),(7038,12046),

107+ (125219574,4),(803694088,1),(107081726,1),(39294,21699),(16361,38191),

108+ (132561123,12),(1760,23499),(847543,484),(175687349,1),(2963,252678),

109+ (6248,224553),(27596,4606),(5422922,121),(1542,485890),(131,583035),

110+ (59096,4925),(3637115,132),(0,947225435),(86854,6794),(2984745,339),

111+ (760129569,1),(1,68260595),(380835652,2),(430575,2579),(54514,7211),

112+ (15550606,3),(9,27367402),(3007053,207),(7060988,60),(28560,27130),

113+ (1355,21087),(10880,53059),(14563646,4),(461886361,1),(2,169260724),

114+ (241454126,2),(406797,1),(61631630,16),(44473,5943),(63869104,12),

115+ (950300,1528),(2113,62333),(120817,9358),(100261456,1),(426764723,1),

116+ (119,12723684),(3,53358711),(4448071,18),(1,230278091),(238,232102),

117+ (8,57316440),(42437979,10),(6769,19555),(48590,22006),(11500585,79),

118+ (2808,97638),(42,26952545),(11,32104194),(23954638,1),(785427272,0),

119+ (513,81379),(31333960,37),(897772,1009),(4,25679692),(103027993,12),

120+ (104972702,11),(546,443401),(7,65137092),(88574269,3),(872139069,0),

121+ (2,97417121),(378802603,0),(141071401,4),(22613,10575),(2191743,118),

122+ (470,116119),(7062,38166),(231056,1847),(43901963,9),(2400,70640),

123+ (63553,1555),(34,11249573),(815174,1820),(997894011,0),(98881794,2),

124+ (5448,43132),(27956,9),(904926,1357),(112608626,3),(124,613021),

125+ (282086,1966),(99,10656881),(113799,1501),(433318,2085),(442,948171),

126+ (165380,1043),(28,14372905),(14880,50462),(2386,219918),(229,1797565),

127+ (1174961,298),(3925,41833),(3903515,299),(15690452,111),(360860521,3),

128+ (7440846,81),(2541026,507),(0,492448477),(6869,82469),(245,8322939),

129+ (3503496,253),(123495298,0),(150963,2299),(33,4408482),(1,200911107),

130+ (305,252121),(13,123369189),(215846,8181),(2440,65387),(776764401,1),

131+ (1241172,434),(8,15493155),(81953961,6),(17884993,5),(26,6893822),

132+ (0,502035190),(1,582451018),(2,514870139),(227,3625619),(49,12720258),

133+ (1456769,207),(94797661,10),(234407,893),(26843,5783),(15688,24547),

134+ (4091,86268),(4339448,151),(21360,6294),(397046497,2),(1227,205936),

135+ (9966,21959),(160046791,1),(0,159992224),(27,24974797),(19177,29334),

136+ (4136148,42),(21179785,53),(61256583,31),(385,344176),(7,11934915),

137+ (1,18992566),(3488065,5),(768021,224),(36288474,7),(8624,117561),

138+ (8,20341439),(5903,261475),(561,1007618),(1738,392327),(633049,1708)]

benchmarks/haskell/Benchmarks/Programs/BigTable.hs+2−2

@@ -11,7 +11,7 @@ module Benchmarks.Programs.BigTable

1111 ( benchmark

1212 ) where

1313

14−import Criterion (Benchmark, bench)

14+import Criterion (Benchmark, bench, whnfIO)

1515import Data.Monoid (mappend, mconcat)

1616import Data.Text.Lazy.Builder (Builder, fromText, toLazyText)

1717import Data.Text.Lazy.IO (hPutStr)

@@ -19,7 +19,7 @@ import System.IO (Handle)

1919import qualified Data.Text as T

2020

2121benchmark :: Handle -> IO Benchmark

22−benchmark sink = return $ bench "BigTable" $ do

22+benchmark sink = return $ bench "BigTable" $ whnfIO $ do

2323 hPutStr sink "Content-Type: text/html\n\n<table>"

2424 hPutStr sink . toLazyText . makeTable =<< rows

2525 hPutStr sink "</table>"

benchmarks/haskell/Benchmarks/Programs/Cut.hs+2−2

@@ -16,7 +16,7 @@ module Benchmarks.Programs.Cut

1616 ( benchmark

1717 ) where

1818

19−import Criterion (Benchmark, bgroup, bench)

19+import Criterion (Benchmark, bgroup, bench, whnfIO)

2020import System.IO (Handle, hPutStr)

2121import qualified Data.ByteString as B

2222import qualified Data.ByteString.Char8 as BC

@@ -40,7 +40,7 @@ benchmark p sink from to = return $ bgroup "Cut"

4040 , bench' "LazyTextByteString" lazyTextByteString

4141 ]

4242 where

43− bench' n s = bench n (s p sink from to)

43+ bench' n s = bench n $ whnfIO (s p sink from to)

4444

4545string :: FilePath -> Handle -> Int -> Int -> IO ()

4646string fp sink from to = do

benchmarks/haskell/Benchmarks/Programs/Fold.hs+2−2

@@ -21,7 +21,7 @@ import Data.List (foldl')

2121import Data.List (intersperse)

2222import Data.Monoid (mempty, mappend, mconcat)

2323import System.IO (Handle)

24−import Criterion (Benchmark, bench)

24+import Criterion (Benchmark, bench, whnfIO)

2525import qualified Data.Text as T

2626import qualified Data.Text.IO as T

2727import qualified Data.Text.Lazy.Builder as TLB

@@ -30,7 +30,7 @@ import qualified Data.Text.Lazy.IO as TL

3030

3131benchmark :: FilePath -> Handle -> IO Benchmark

3232benchmark i o = return $

33− bench "Fold" $ T.readFile i >>= TL.hPutStr o . fold 80

33+ bench "Fold" $ whnfIO $ T.readFile i >>= TL.hPutStr o . fold 80

3434

3535-- | We represent a paragraph by a word list

3636--

benchmarks/haskell/Benchmarks/Programs/Sort.hs+10−9

@@ -17,7 +17,7 @@ module Benchmarks.Programs.Sort

1717 ( benchmark

1818 ) where

1919

20−import Criterion (Benchmark, bgroup, bench)

20+import Criterion (Benchmark, bgroup, bench, whnfIO)

2121import Data.Monoid (mconcat)

2222import System.IO (Handle, hPutStr)

2323import qualified Data.ByteString as B

@@ -35,16 +35,17 @@ import qualified Data.Text.Lazy.IO as TL

3535

3636benchmark :: FilePath -> Handle -> IO Benchmark

3737benchmark i o = return $ bgroup "Sort"

38− [ bench "String" $ readFile i >>= hPutStr o . string

39− , bench "ByteString" $ B.readFile i >>= B.hPutStr o . byteString

40− , bench "LazyByteString" $ BL.readFile i >>= BL.hPutStr o . lazyByteString

41− , bench "Text" $ T.readFile i >>= T.hPutStr o . text

42− , bench "LazyText" $ TL.readFile i >>= TL.hPutStr o . lazyText

43− , bench "TextByteString" $ B.readFile i >>=

38+ [ bench "String" $ whnfIO $ readFile i >>= hPutStr o . string

39+ , bench "ByteString" $ whnfIO $ B.readFile i >>= B.hPutStr o . byteString

40+ , bench "LazyByteString" $ whnfIO $

41+ BL.readFile i >>= BL.hPutStr o . lazyByteString

42+ , bench "Text" $ whnfIO $ T.readFile i >>= T.hPutStr o . text

43+ , bench "LazyText" $ whnfIO $ TL.readFile i >>= TL.hPutStr o . lazyText

44+ , bench "TextByteString" $ whnfIO $ B.readFile i >>=

4445 B.hPutStr o . T.encodeUtf8 . text . T.decodeUtf8

45− , bench "LazyTextByteString" $ BL.readFile i >>=

46+ , bench "LazyTextByteString" $ whnfIO $ BL.readFile i >>=

4647 BL.hPutStr o . TL.encodeUtf8 . lazyText . TL.decodeUtf8

47− , bench "TextBuilder" $ B.readFile i >>=

48+ , bench "TextBuilder" $ whnfIO $ B.readFile i >>=

4849 BL.hPutStr o . TL.encodeUtf8 . textBuilder . T.decodeUtf8

4950 ]

5051

benchmarks/haskell/Benchmarks/Programs/StripTags.hs+5−5

@@ -15,7 +15,7 @@ module Benchmarks.Programs.StripTags

1515 ( benchmark

1616 ) where

1717

18−import Criterion (Benchmark, bgroup, bench)

18+import Criterion (Benchmark, bgroup, bench, whnfIO)

1919import Data.List (mapAccumL)

2020import System.IO (Handle, hPutStr)

2121import qualified Data.ByteString as B

@@ -26,10 +26,10 @@ import qualified Data.Text.IO as T

2626

2727benchmark :: FilePath -> Handle -> IO Benchmark

2828benchmark i o = return $ bgroup "StripTags"

29− [ bench "String" $ readFile i >>= hPutStr o . string

30− , bench "ByteString" $ B.readFile i >>= B.hPutStr o . byteString

31− , bench "Text" $ T.readFile i >>= T.hPutStr o . text

32− , bench "TextByteString" $

29+ [ bench "String" $ whnfIO $ readFile i >>= hPutStr o . string

30+ , bench "ByteString" $ whnfIO $ B.readFile i >>= B.hPutStr o . byteString

31+ , bench "Text" $ whnfIO $ T.readFile i >>= T.hPutStr o . text

32+ , bench "TextByteString" $ whnfIO $

3333 B.readFile i >>= B.hPutStr o . T.encodeUtf8 . text . T.decodeUtf8

3434 ]

3535

benchmarks/haskell/Benchmarks/Programs/Throughput.hs+8−8

@@ -18,7 +18,7 @@ module Benchmarks.Programs.Throughput

1818 ( benchmark

1919 ) where

2020

21−import Criterion (Benchmark, bgroup, bench)

21+import Criterion (Benchmark, bgroup, bench, whnfIO)

2222import System.IO (Handle, hPutStr)

2323import qualified Data.ByteString as B

2424import qualified Data.ByteString.Lazy as BL

@@ -29,13 +29,13 @@ import qualified Data.Text.Lazy.IO as TL

2929

3030benchmark :: FilePath -> Handle -> IO Benchmark

3131benchmark fp sink = return $ bgroup "Throughput"

32− [ bench "String" $ readFile fp >>= hPutStr sink

33− , bench "ByteString" $ B.readFile fp >>= B.hPutStr sink

34− , bench "LazyByteString" $ BL.readFile fp >>= BL.hPutStr sink

35− , bench "Text" $ T.readFile fp >>= T.hPutStr sink

36− , bench "LazyText" $ TL.readFile fp >>= TL.hPutStr sink

37− , bench "TextByteString" $

32+ [ bench "String" $ whnfIO $ readFile fp >>= hPutStr sink

33+ , bench "ByteString" $ whnfIO $ B.readFile fp >>= B.hPutStr sink

34+ , bench "LazyByteString" $ whnfIO $ BL.readFile fp >>= BL.hPutStr sink

35+ , bench "Text" $ whnfIO $ T.readFile fp >>= T.hPutStr sink

36+ , bench "LazyText" $ whnfIO $ TL.readFile fp >>= TL.hPutStr sink

37+ , bench "TextByteString" $ whnfIO $

3838 B.readFile fp >>= B.hPutStr sink . T.encodeUtf8 . T.decodeUtf8

39− , bench "LazyTextByteString" $

39+ , bench "LazyTextByteString" $ whnfIO $

4040 BL.readFile fp >>= BL.hPutStr sink . TL.encodeUtf8 . TL.decodeUtf8

4141 ]

benchmarks/haskell/Benchmarks/Pure.hs+7

@@ -262,6 +262,13 @@ benchmark kind fp = do

262262 , benchBSL $ nf (BL.map toUpper) bla

263263 , benchS $ nf (L.map toUpper) sa

264264 ]

265+ , bgroup "uncons"

266+ [ benchT $ nf T.uncons ta

267+ , benchTL $ nf TL.uncons tla

268+ , benchBS $ nf BS.uncons bsa

269+ , benchBSL $ nf BL.uncons bla

270+ , benchS $ nf L.uncons sa

271+ ]

265272 , bgroup "words"

266273 [ benchT $ nf T.words ta

267274 , benchTL $ nf TL.words tla

benchmarks/haskell/Benchmarks/ReadNumbers.hs+2−5

@@ -25,8 +25,7 @@ import Data.List (foldl')

2525import Numeric (readDec, readFloat, readHex)

2626import qualified Data.ByteString.Char8 as B

2727import qualified Data.ByteString.Lazy.Char8 as BL

28−import qualified Data.ByteString.Lex.Double as B

29−import qualified Data.ByteString.Lex.Lazy.Double as BL

28+import qualified Data.ByteString.Lex.Fractional as B

3029import qualified Data.Text as T

3130import qualified Data.Text.IO as T

3231import qualified Data.Text.Lazy as TL

@@ -63,12 +62,10 @@ benchmark fp = do

6362 whnf (double . text TL.rational) tl

6463

6564 , bench "DecimalByteString" $ whnf (int . byteString B.readInt) b

66− , bench "DoubleByteString" $ whnf (double . byteString B.readDouble) b

65+ , bench "DoubleByteString" $ whnf (double . byteString B.readDecimal) b

6766

6867 , bench "DecimalLazyByteString" $

6968 whnf (int . byteString BL.readInt) bl

70− , bench "DoubleLazyByteString" $

71− whnf (double . byteString BL.readDouble) bl

7269 ]

7370 where

7471 -- Used for fixing types

benchmarks/haskell/Benchmarks/Replace.hs−1

@@ -28,7 +28,6 @@ benchmark fp pat sub = do

2828 !b = T.encodeUtf8 t

2929 return $ bgroup "Replace" [

3030 bench "Text" $ nf (T.length . T.replace tpat tsub) t

31− , bench "Text2" $ nf (T.length . T.replace2 tpat tsub) t

3231 , bench "ByteString" $ nf (BL.length . B.replace bpat bsub) b

3332 , bench "LazyText" $ nf (TL.length . TL.replace tlpat tlsub) tl

3433 , bench "LazyByteString" $ nf (BL.length . BL.replace blpat blsub) bl

benchmarks/haskell/Benchmarks/Stream.hs+20−9

@@ -13,16 +13,16 @@ module Benchmarks.Stream

1313

1414import Control.DeepSeq (NFData (..))

1515import Criterion (Benchmark, bgroup, bench, nf)

16−import Data.Text.Fusion.Internal (Step (..), Stream (..))

16+import Data.Text.Internal.Fusion.Types (Step (..), Stream (..))

1717import qualified Data.Text.Encoding as T

1818import qualified Data.Text.Encoding.Error as E

19−import qualified Data.Text.Encoding.Fusion as T

20−import qualified Data.Text.Encoding.Fusion.Common as F

21−import qualified Data.Text.Fusion as T

19+import qualified Data.Text.Internal.Encoding.Fusion as T

20+import qualified Data.Text.Internal.Encoding.Fusion.Common as F

21+import qualified Data.Text.Internal.Fusion as F

2222import qualified Data.Text.IO as T

2323import qualified Data.Text.Lazy.Encoding as TL

24−import qualified Data.Text.Lazy.Encoding.Fusion as TL

25−import qualified Data.Text.Lazy.Fusion as TL

24+import qualified Data.Text.Internal.Lazy.Encoding.Fusion as TL

25+import qualified Data.Text.Internal.Lazy.Fusion as FL

2626import qualified Data.Text.Lazy.IO as TL

2727

2828instance NFData a => NFData (Stream a) where

@@ -53,14 +53,25 @@ benchmark fp = do

5353 !utf32beL = TL.encodeUtf32BE tl

5454

5555 -- For the functions which operate on streams

56− let !s = T.stream t

56+ let !s = F.stream t

5757

5858 return $ bgroup "Stream"

5959

6060 -- Fusion

6161 [ bgroup "stream" $

62− [ bench "Text" $ nf T.stream t

63− , bench "LazyText" $ nf TL.stream tl

62+ [ bench "Text" $ nf F.stream t

63+ , bench "LazyText" $ nf FL.stream tl

64+ ]

65+ -- must perform exactly the same as stream above due to

66+ -- stream/unstream (i.e. stream after unstream) fusion

67+ , bgroup "stream-fusion" $

68+ [ bench "Text" $ nf (F.stream . F.unstream . F.stream) t

69+ , bench "LazyText" $ nf (FL.stream . FL.unstream . FL.stream) tl

70+ ]

71+ -- measure the overhead of unstream after stream

72+ , bgroup "stream-unstream" $

73+ [ bench "Text" $ nf (F.unstream . F.stream) t

74+ , bench "LazyText" $ nf (FL.unstream . FL.stream) tl

6475 ]

6576

6677 -- Encoding.Fusion

benchmarks/text-benchmarks.cabal+18−6

@@ -3,7 +3,7 @@ version: 0.0.0.0

33synopsis: Benchmarks for the text package

44description: Benchmarks for the text package

55homepage: https://bitbucket.org/bos/text

6−license: BSD3

6+license: BSD2

77license-file: ../LICENSE

88author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

99 Bryan O'Sullivan <bos@serpentine.com>,

@@ -15,9 +15,15 @@ build-type: Simple

1515

1616cabal-version: >=1.2

1717

18+flag bytestring-builder

19+ description: Depend on the bytestring-builder package for backwards compatibility.

20+ default: False

21+ manual: False

22+

1823flag llvm

1924 description: use LLVM

2025 default: False

26+ manual: True

2127

2228executable text-benchmarks

2329 hs-source-dirs: haskell ..

@@ -25,24 +31,30 @@ executable text-benchmarks

2531 cbits/time_iconv.c

2632 include-dirs: ../include

2733 main-is: Benchmarks.hs

28− ghc-options: -Wall -O2

34+ ghc-options: -Wall -O2 -rtsopts

2935 if flag(llvm)

3036 ghc-options: -fllvm

3137 cpp-options: -DHAVE_DEEPSEQ -DINTEGER_GMP

3238 build-depends: base == 4.*,

3339 binary,

3440 blaze-builder,

35− bytestring,

36− bytestring-lexing,

41+ bytestring-lexing >= 0.5.0,

3742 containers,

38− criterion >= 0.6.0.1,

43+ criterion >= 0.10.0.0,

3944 deepseq,

4045 directory,

4146 filepath,

4247 ghc-prim,

4348 integer-gmp,

4449 stringsearch,

45− utf8-string

50+ utf8-string,

51+ vector

52+

53+ if flag(bytestring-builder)

54+ build-depends: bytestring >= 0.9 && < 0.10.4,

55+ bytestring-builder >= 0.10.4

56+ else

57+ build-depends: bytestring >= 0.10.4

4658

4759executable text-multilang

4860 hs-source-dirs: haskell

cbits/cbits.c+113−12

@@ -68,8 +68,8 @@ decode(uint32_t *state, uint32_t* codep, uint32_t byte) {

6868 * an UTF16 array

6969 */

7070void

71−_hs_text_decode_latin1(uint16_t *dest, const uint8_t const *src,

72− const uint8_t const *srcend)

71+_hs_text_decode_latin1(uint16_t *dest, const uint8_t *src,

72+ const uint8_t *srcend)

7373{

7474 const uint8_t *p = src;

7575

@@ -127,11 +127,18 @@ _hs_text_decode_latin1(uint16_t *dest, const uint8_t const *src,

127127 * state0 != UTF8_ACCEPT, UTF8_REJECT

128128 *

129129 */

130−const uint8_t *

131−_hs_text_decode_utf8_state(uint16_t *const dest, size_t *destoff,

132− const uint8_t **const src,

133− const uint8_t *const srcend,

134− uint32_t *codepoint0, uint32_t *state0)

130+#if defined(__GNUC__) || defined(__clang__)

131+static inline uint8_t const *

132+_hs_text_decode_utf8_int(uint16_t *const dest, size_t *destoff,

133+ const uint8_t **src, const uint8_t *srcend,

134+ uint32_t *codepoint0, uint32_t *state0)

135+ __attribute((always_inline));

136+#endif

137+

138+static inline uint8_t const *

139+_hs_text_decode_utf8_int(uint16_t *const dest, size_t *destoff,

140+ const uint8_t **src, const uint8_t *srcend,

141+ uint32_t *codepoint0, uint32_t *state0)

135142{

136143 uint16_t *d = dest + *destoff;

137144 const uint8_t *s = *src, *last = *src;

@@ -185,10 +192,6 @@ _hs_text_decode_utf8_state(uint16_t *const dest, size_t *destoff,

185192 last = s;

186193 }

187194

188− /* Invalid encoding, back up to the errant character */

189− if (state == UTF8_REJECT)

190− s -= 1;

191−

192195 *destoff = d - dest;

193196 *codepoint0 = codepoint;

194197 *state0 = state;

@@ -197,6 +200,19 @@ _hs_text_decode_utf8_state(uint16_t *const dest, size_t *destoff,

197200 return s;

198201}

199202

203+uint8_t const *

204+_hs_text_decode_utf8_state(uint16_t *const dest, size_t *destoff,

205+ const uint8_t **src,

206+ const uint8_t *srcend,

207+ uint32_t *codepoint0, uint32_t *state0)

208+{

209+ uint8_t const *ret = _hs_text_decode_utf8_int(dest, destoff, src, srcend,

210+ codepoint0, state0);

211+ if (*state0 == UTF8_REJECT)

212+ ret -=1;

213+ return ret;

214+}

215+

200216/*

201217 * Helper to decode buffer and discard final decoder state

202218 */

@@ -206,5 +222,90 @@ _hs_text_decode_utf8(uint16_t *const dest, size_t *destoff,

206222{

207223 uint32_t codepoint;

208224 uint32_t state = UTF8_ACCEPT;

209− return _hs_text_decode_utf8_state(dest, destoff, &src, srcend, &codepoint, &state);

225+ uint8_t const *ret = _hs_text_decode_utf8_int(dest, destoff, &src, srcend,

226+ &codepoint, &state);

227+ /* Back up if we have an incomplete or invalid encoding */

228+ if (state != UTF8_ACCEPT)

229+ ret -= 1;

230+ return ret;

231+}

232+

233+void

234+_hs_text_encode_utf8(uint8_t **destp, const uint16_t *src, size_t srcoff,

235+ size_t srclen)

236+{

237+ const uint16_t *srcend;

238+ uint8_t *dest = *destp;

239+

240+ src += srcoff;

241+ srcend = src + srclen;

242+

243+ ascii:

244+#if defined(__x86_64__)

245+ while (srcend - src >= 4) {

246+ uint64_t w = *((uint64_t *) src);

247+

248+ if (w & 0xFF80FF80FF80FF80ULL) {

249+ if (!(w & 0x000000000000FF80ULL)) {

250+ *dest++ = w & 0xFFFF;

251+ src++;

252+ if (!(w & 0x00000000FF800000ULL)) {

253+ *dest++ = (w >> 16) & 0xFFFF;

254+ src++;

255+ if (!(w & 0x0000FF8000000000ULL)) {

256+ *dest++ = (w >> 32) & 0xFFFF;

257+ src++;

258+ }

259+ }

260+ }

261+ break;

262+ }

263+ *dest++ = w & 0xFFFF;

264+ *dest++ = (w >> 16) & 0xFFFF;

265+ *dest++ = (w >> 32) & 0xFFFF;

266+ *dest++ = w >> 48;

267+ src += 4;

268+ }

269+#endif

270+

271+#if defined(__i386__)

272+ while (srcend - src >= 2) {

273+ uint32_t w = *((uint32_t *) src);

274+

275+ if (w & 0xFF80FF80)

276+ break;

277+ *dest++ = w & 0xFFFF;

278+ *dest++ = w >> 16;

279+ src += 2;

280+ }

281+#endif

282+

283+ while (src < srcend) {

284+ uint16_t w = *src++;

285+

286+ if (w <= 0x7F) {

287+ *dest++ = w;

288+ /* An ASCII byte is likely to begin a run of ASCII bytes.

289+ Falling back into the fast path really helps performance. */

290+ goto ascii;

291+ }

292+ else if (w <= 0x7FF) {

293+ *dest++ = (w >> 6) | 0xC0;

294+ *dest++ = (w & 0x3f) | 0x80;

295+ }

296+ else if (w < 0xD800 || w > 0xDBFF) {

297+ *dest++ = (w >> 12) | 0xE0;

298+ *dest++ = ((w >> 6) & 0x3F) | 0x80;

299+ *dest++ = (w & 0x3F) | 0x80;

300+ } else {

301+ uint32_t c = ((((uint32_t) w) - 0xD800) << 10) +

302+ (((uint32_t) *src++) - 0xDC00) + 0x10000;

303+ *dest++ = (c >> 18) | 0xF0;

304+ *dest++ = ((c >> 12) & 0x3F) | 0x80;

305+ *dest++ = ((c >> 6) & 0x3F) | 0x80;

306+ *dest++ = (c & 0x3F) | 0x80;

307+ }

308+ }

309+

310+ *destp = dest;

210311}

changelog.mdnew file+164

@@ -0,0 +1,164 @@

1+1.2.2.1

3+* The switch to `integer-pure` in 1.2.2.0 was apparently mistaken.

4+ The build flag has been renamed accordingly. Your army of diligent

5+ maintainers apologizes for the churn.

7+* Spec compliance: toCaseFold now follows the Unicode 8.0 spec

8+ (updated from 7.0)

10+* An STG lint error has been fixed

11+

12+1.2.2.0

13+

14+* The `integer-simple` package, upon which this package optionally

15+ depended, has been replaced with `integer-pure`. The build flag has

16+ been renamed accordingly.

17+

18+* Bug fix: For the `Binary` instance, If UTF-8 decoding fails during a

19+ `get`, the error is propagated via `fail` instead of an uncatchable

20+ crash.

21+

22+* New function: `takeWhileEnd`

23+

24+* New instances for the `Text` types:

25+ * if `base` >= 4.7: `PrintfArg`

26+ * if `base` >= 4.9: `Semigroup`

27+

28+1.2.1.3

29+

30+* Bug fix: As it turns out, moving the literal rewrite rules to simplifier

31+ phase 2 does not prevent competition with the `unpack` rule, which is

32+ also active in this phase. Unfortunately this was hidden due to a silly

33+ test environment mistake. Moving literal rules back to phase 1 finally

34+ fixes GHC Trac #10528 correctly.

35+

36+1.2.1.2

37+

38+* Bug fix: Run literal rewrite rules in simplifier phase 2.

39+ The behavior of the simplifier changed in GHC 7.10.2,

40+ causing these rules to fail to fire, leading to poor code generation

41+ and long compilation times. See

42+ [GHC Trac #10528](https://ghc.haskell.org/trac/ghc/ticket/10528).

43+

44+1.2.1.1

45+

46+* Expose unpackCString#, which you should never use.

47+

48+1.2.1.0

49+

50+* Added Binary instances for both Text types. (If you have previously

51+ been using the text-binary package to get a Binary instance, it is

52+ now obsolete.)

53+

54+1.2.0.6

55+

56+* Fixed a space leak in UTF-8 decoding

57+

58+1.2.0.5

59+

60+* Feature parity: repeat, cycle, iterate are now implemented for lazy

61+ Text, and the Data instance is more complete

62+

63+* Build speed: an inliner space explosion has been fixed with toCaseFold

64+

65+* Bug fix: encoding Int to a Builder would infinite-loop if the

66+ integer-simple package was used

67+

68+* Deprecation: OnEncodeError and EncodeError are deprecated, as they

69+ are never used

70+

71+* Internals: some types that are used internally in fusion-related

72+ functions have moved around, been renamed, or been deleted (we don't

73+ bump the major version if .Internal modules change)

74+

75+* Spec compliance: toCaseFold now follows the Unicode 7.0 spec

76+ (updated from 6.3)

77+

78+1.2.0.4

79+

80+* Fixed an incompatibility with base < 4.5

81+

82+1.2.0.3

83+

84+* Update formatRealFloat to correspond to the definition in versions

85+ of base newer than 4.5 (https://github.com/bos/text/issues/105)

86+

87+1.2.0.2

88+

89+* Bumped lower bound on deepseq to 1.4 for compatibility with the

90+ upcoming GHC 7.10

91+

92+1.2.0.1

93+

94+* Fixed a buffer overflow in rendering of large Integers

95+ (https://github.com/bos/text/issues/99)

96+

97+1.2.0.0

98+

99+* Fixed an integer overflow in the replace function

100+ (https://github.com/bos/text/issues/81)

101+

102+* Fixed a hang in lazy decodeUtf8With

103+ (https://github.com/bos/text/issues/87)

104+

105+* Reduced codegen bloat caused by use of empty and single-character

106+ literals

107+

108+* Added an instance of IsList for GHC 7.8 and above

109+

110+1.1.1.0

111+

112+* The Data.Data instance now allows gunfold to work, via a virtual

113+ pack constructor

114+

115+* dropEnd, takeEnd: new functions

116+

117+* Comparing the length of a Text against a number can now

118+ short-circuit in more cases

119+

120+1.1.0.1

121+

122+* streamDecodeUtf8: fixed gh-70, did not return all unconsumed bytes

123+ in single-byte chunks

124+

125+1.1.0.0

126+

127+* encodeUtf8: Performance is improved by up to 4x.

128+

129+* encodeUtf8Builder, encodeUtf8BuilderEscaped: new functions,

130+ available only if bytestring >= 0.10.4.0 is installed, that allow

131+ very fast and flexible encoding of a Text value to a bytestring

132+ Builder.

133+

134+ As an example of the performance gain to be had, the

135+ encodeUtf8BuilderEscaped function helps to double the speed of JSON

136+ encoding in the latest version of aeson! (Note: if all you need is a

137+ plain ByteString, encodeUtf8 is still the faster way to go.)

138+

139+* All of the internal module hierarchy is now publicly exposed. If a

140+ module is in the .Internal hierarchy, or is documented as internal,

141+ use at your own risk - there are no API stability guarantees for

142+ internal modules!

143+

144+1.0.0.1

145+

146+* decodeUtf8: Fixed a regression that caused us to incorrectly

147+ identify truncated UTF-8 as valid (gh-61)

148+

149+1.0.0.0

150+

151+* Added support for Unicode 6.3.0 to case conversion functions

152+

153+* New function toTitle converts words in a string to title case

154+

155+* New functions peekCStringLen and withCStringLen simplify

156+ interoperability with C functionns

157+

158+* Added support for decoding UTF-8 in stream-friendly fashion

159+

160+* Fixed a bug in mapAccumL

161+

162+* Added trusted Haskell support

163+

164+* Removed support for GHC 6.10 (released in 2008) and older

scripts/CaseFolding.hs+1−1

@@ -37,7 +37,7 @@ mapCF :: CaseFolding -> [String]

3737mapCF (CF _ ms) = typ ++ (map nice . filter p $ ms) ++ [last]

3838 where

3939 typ = ["foldMapping :: forall s. Char -> s -> Step (CC s) Char"

40− ,"{-# INLINE foldMapping #-}"]

40+ ,"{-# NOINLINE foldMapping #-}"]

4141 last = "foldMapping c s = Yield (toLower c) (CC s '\\0' '\\0')"

4242 nice c = "-- " ++ name c ++ "\n" ++

4343 "foldMapping " ++ showC (code c) ++ " s = Yield " ++ x ++ " (CC s " ++ y ++ " " ++ z ++ ")"

scripts/CaseMapping.hs+4−4

@@ -8,7 +8,7 @@ import SpecialCasing

88main = do

99 args <- getArgs

1010 let oname = case args of

11− [] -> "../Data/Text/Fusion/CaseMapping.hs"

11+ [] -> "../Data/Text/Internal/Fusion/CaseMapping.hs"

1212 [o] -> o

1313 psc <- parseSC "SpecialCasing.txt"

1414 pcf <- parseCF "CaseFolding.txt"

@@ -24,12 +24,12 @@ main = do

2424 mapM_ (hPutStrLn h) $

2525 ["{-# LANGUAGE Rank2Types #-}"

2626 ,"-- AUTOMATICALLY GENERATED - DO NOT EDIT"

27− ,"-- Generated by scripts/SpecialCasing.hs"] ++

27+ ,"-- Generated by scripts/CaseMapping.hs"] ++

2828 comments ++

2929 [""

30− ,"module Data.Text.Fusion.CaseMapping where"

30+ ,"module Data.Text.Internal.Fusion.CaseMapping where"

3131 ,"import Data.Char"

32− ,"import Data.Text.Fusion.Internal"

32+ ,"import Data.Text.Internal.Fusion.Types"

3333 ,""]

3434 mapM_ (hPutStrLn h) (mapSC "upper" upper toUpper scs)

3535 mapM_ (hPutStrLn h) (mapSC "lower" lower toLower scs)

scripts/SpecialCasing.hs+1−1

@@ -43,7 +43,7 @@ mapSC which access twiddle (SC _ ms) =

4343 typ ++ (map nice . filter p $ ms) ++ [last]

4444 where

4545 typ = [which ++ "Mapping :: forall s. Char -> s -> Step (CC s) Char"

46− ,"{-# INLINE " ++ which ++ "Mapping #-}"]

46+ ,"{-# NOINLINE " ++ which ++ "Mapping #-}"]

4747 last = which ++ "Mapping c s = Yield (to" ++ ucFirst which ++ " c) (CC s '\\0' '\\0')"

4848 nice c = "-- " ++ name c ++ "\n" ++

4949 which ++ "Mapping " ++ showC (code c) ++ " s = Yield " ++ x ++ " (CC s " ++ y ++ " " ++ z ++ ")"

tests/LiteralRuleTest.hsnew file+21

@@ -0,0 +1,21 @@

1+{-# LANGUAGE OverloadedStrings #-}

3+module LiteralRuleTest where

5+import Data.Text (Text)

7+-- This should produce 8 firings of the "TEXT literal" rule

8+strings :: [Text]

9+strings = [ "abstime", "aclitem", "bit", "bool", "box", "bpchar", "bytea", "char" ]

10+

11+-- This should produce 7 firings of the "TEXT literal UTF8" rule

12+utf8Strings :: [Text]

13+utf8Strings = [ "\0abstime", "\0aclitem", "\xfefe bit", "\0bool", "\0box", "\0bpchar", "\0bytea" ]

14+

15+-- This should produce 4 firings of the "TEXT empty literal" rule

16+empties :: [Text]

17+empties = [ "", "", "", "" ]

18+

19+-- This should produce 5 firings of the "TEXT empty literal" rule

20+--singletons :: [Text]

21+--singletons = [ "a", "b", "c", "d", "e" ]

tests/Makefile+6−1

@@ -1,5 +1,10 @@

11count = 1000

3+all: coverage literal-rule-test

5+literal-rule-test:

6+ ./literal-rule-test.sh

38coverage: build coverage/hpc_index.html

510build: text-test-data

@@ -32,4 +37,4 @@ coverage/hpc_index.html: coverage/coverage.tix

3237clean:

3338 rm -rf dist coverage .hpc

3439

35−.PHONY: build coverage

40+.PHONY: build coverage all literal-rule-test

tests/Tests/Properties.hs+228−93

@@ -1,64 +1,61 @@

1−-- | General quicktest properties for the text library

2−--

3−{-# LANGUAGE BangPatterns, CPP, FlexibleInstances, OverloadedStrings,

1+-- | QuickCheck properties for the text library.

3+{-# LANGUAGE BangPatterns, FlexibleInstances, OverloadedStrings,

44 ScopedTypeVariables, TypeSynonymInstances #-}

5−{-# OPTIONS_GHC -fno-enable-rewrite-rules #-}

6−{-# OPTIONS_GHC -fno-warn-missing-signatures #-}

5+{-# OPTIONS_GHC -fno-enable-rewrite-rules -fno-warn-missing-signatures #-}

76module Tests.Properties

87 (

98 tests

109 ) where

1110

12−import Test.QuickCheck

13−import Test.QuickCheck.Monadic

14−import Text.Show.Functions ()

15−

11+import Control.Applicative ((<$>), (<*>))

1612import Control.Arrow ((***), second)

17−import Control.Exception (catch)

13+import Data.Bits ((.&.))

1814import Data.Char (chr, isDigit, isHexDigit, isLower, isSpace, isUpper, ord)

1915import Data.Int (Int8, Int16, Int32, Int64)

2016import Data.Monoid (Monoid(..))

21−import Data.String (fromString)

17+import Data.String (IsString(fromString))

2218import Data.Text.Encoding.Error

2319import Data.Text.Foreign

24−import Data.Text.Fusion.Size

20+import Data.Text.Internal.Encoding.Utf8

21+import Data.Text.Internal.Fusion.Size

22+import Data.Text.Internal.Search (indices)

2523import Data.Text.Lazy.Read as TL

2624import Data.Text.Read as T

27−import Data.Text.Search (indices)

2825import Data.Word (Word, Word8, Word16, Word32, Word64)

29−import Numeric (showHex)

26+import Numeric (showEFloat, showFFloat, showGFloat, showHex)

27+import Prelude hiding (replicate)

3028import Test.Framework (Test, testGroup)

3129import Test.Framework.Providers.QuickCheck2 (testProperty)

30+import Test.QuickCheck hiding ((.&.))

31+import Test.QuickCheck.Monadic

32+import Test.QuickCheck.Property (Property(..))

33+import Tests.QuickCheckUtils

34+import Tests.Utils

35+import Text.Show.Functions ()

36+import qualified Control.Exception as Exception

3237import qualified Data.Bits as Bits (shiftL, shiftR)

3338import qualified Data.ByteString as B

3439import qualified Data.ByteString.Lazy as BL

3540import qualified Data.List as L

3641import qualified Data.Text as T

3742import qualified Data.Text.Encoding as E

38−import qualified Data.Text.Fusion as S

39−import qualified Data.Text.Fusion.Common as S

4043import qualified Data.Text.IO as T

44+import qualified Data.Text.Internal.Fusion as S

45+import qualified Data.Text.Internal.Fusion.Common as S

46+import qualified Data.Text.Internal.Lazy.Fusion as SL

47+import qualified Data.Text.Internal.Lazy.Search as S (indices)

48+import qualified Data.Text.Internal.Unsafe.Shift as U

4149import qualified Data.Text.Lazy as TL

4250import qualified Data.Text.Lazy.Builder as TB

4351import qualified Data.Text.Lazy.Builder.Int as TB

4452import qualified Data.Text.Lazy.Builder.RealFloat as TB

4553import qualified Data.Text.Lazy.Encoding as EL

46−import qualified Data.Text.Lazy.Fusion as SL

4754import qualified Data.Text.Lazy.IO as TL

48−import qualified Data.Text.Lazy.Search as S (indices)

49−import qualified Data.Text.UnsafeShift as U

5055import qualified System.IO as IO

51−

52−import Tests.QuickCheckUtils

53−import Tests.Utils

56+import qualified Tests.Properties.Mul as Mul

5457import qualified Tests.SlowFunctions as Slow

5558

56−#if MIN_VERSION_base(4,6,0)

57−import Prelude hiding (replicate)

58−#else

59−import Prelude hiding (catch, replicate)

60−#endif

61−

6259t_pack_unpack = (T.unpack . T.pack) `eq` id

6360tl_pack_unpack = (TL.unpack . TL.pack) `eq` id

6461t_stream_unstream = (S.unstream . S.stream) `eq` id

@@ -98,35 +95,88 @@ tl_utf32LE = forAll genUnicode $ (EL.decodeUtf32LE . EL.encodeUtf32LE) `eq` id

9895t_utf32BE = forAll genUnicode $ (E.decodeUtf32BE . E.encodeUtf32BE) `eq` id

9996tl_utf32BE = forAll genUnicode $ (EL.decodeUtf32BE . EL.encodeUtf32BE) `eq` id

10097

101−t_utf8_incr = do

102− Positive n <- arbitrary

103− forAll genUnicode $ recode n `eq` id

104− where recode n = T.concat . feedChunksOf n E.streamDecodeUtf8 . E.encodeUtf8

105− feedChunksOf :: Int -> (B.ByteString -> E.Decoding) -> B.ByteString

106− -> [T.Text]

107− feedChunksOf n f bs

108− | B.null bs = []

109− | otherwise = let (a,b) = B.splitAt n bs

110− E.Some t _ f' = f a

111− in t : feedChunksOf n f' b

112−

113−-- This is a poor attempt to ensure that the error handling paths on

114−-- decode are exercised in some way. Proper testing would be rather

115−-- more involved.

116−t_utf8_err :: DecodeErr -> B.ByteString -> Property

117−t_utf8_err (DE _ de) bs = monadicIO $ do

118− l <- run $ let len = T.length (E.decodeUtf8With de bs)

119− in (len `seq` return (Right len)) `catch`

120− (\(e::UnicodeException) -> return (Left e))

121− case l of

122− Left err -> assert $ length (show err) >= 0

123− Right n -> assert $ n >= 0

98+t_utf8_incr = forAll genUnicode $ \s (Positive n) -> (recode n `eq` id) s

99+ where recode n = T.concat . map fst . feedChunksOf n E.streamDecodeUtf8 .

100+ E.encodeUtf8

101+

102+feedChunksOf :: Int -> (B.ByteString -> E.Decoding) -> B.ByteString

103+ -> [(T.Text, B.ByteString)]

104+feedChunksOf n f bs

105+ | B.null bs = []

106+ | otherwise = let (x,y) = B.splitAt n bs

107+ E.Some t b f' = f x

108+ in (t,b) : feedChunksOf n f' y

109+

110+t_utf8_undecoded = forAll genUnicode $ \t ->

111+ let b = E.encodeUtf8 t

112+ ls = concatMap (leftover . E.encodeUtf8 . T.singleton) . T.unpack $ t

113+ leftover = (++ [B.empty]) . init . tail . B.inits

114+ in (map snd . feedChunksOf 1 E.streamDecodeUtf8) b == ls

115+

116+data Badness = Solo | Leading | Trailing

117+ deriving (Eq, Show)

118+

119+instance Arbitrary Badness where

120+ arbitrary = elements [Solo, Leading, Trailing]

121+

122+t_utf8_err :: Badness -> DecodeErr -> Property

123+t_utf8_err bad de = do

124+ let gen = case bad of

125+ Solo -> genInvalidUTF8

126+ Leading -> B.append <$> genInvalidUTF8 <*> genUTF8

127+ Trailing -> B.append <$> genUTF8 <*> genInvalidUTF8

128+ genUTF8 = E.encodeUtf8 <$> genUnicode

129+ forAll gen $ \bs -> MkProperty $ do

130+ onErr <- genDecodeErr de

131+ unProperty . monadicIO $ do

132+ l <- run $ let len = T.length (E.decodeUtf8With onErr bs)

133+ in (len `seq` return (Right len)) `Exception.catch`

134+ (\(e::UnicodeException) -> return (Left e))

135+ assert $ case l of

136+ Left err -> length (show err) >= 0

137+ Right _ -> de /= Strict

124138

125139t_utf8_err' :: B.ByteString -> Property

126140t_utf8_err' bs = monadicIO . assert $ case E.decodeUtf8' bs of

127141 Left err -> length (show err) >= 0

128142 Right t -> T.length t >= 0

129143

144+genInvalidUTF8 :: Gen B.ByteString

145+genInvalidUTF8 = B.pack <$> oneof [

146+ -- invalid leading byte of a 2-byte sequence

147+ (:) <$> choose (0xC0, 0xC1) <*> upTo 1 contByte

148+ -- invalid leading byte of a 4-byte sequence

149+ , (:) <$> choose (0xF5, 0xFF) <*> upTo 3 contByte

150+ -- 4-byte sequence greater than U+10FFFF

151+ , do k <- choose (0x11, 0x13)

152+ let w0 = 0xF0 + (k `Bits.shiftR` 2)

153+ w1 = 0x80 + ((k .&. 3) `Bits.shiftL` 4)

154+ ([w0,w1]++) <$> vectorOf 2 contByte

155+ -- continuation bytes without a start byte

156+ , listOf1 contByte

157+ -- short 2-byte sequence

158+ , (:[]) <$> choose (0xC2, 0xDF)

159+ -- short 3-byte sequence

160+ , (:) <$> choose (0xE0, 0xEF) <*> upTo 1 contByte

161+ -- short 4-byte sequence

162+ , (:) <$> choose (0xF0, 0xF4) <*> upTo 2 contByte

163+ -- overlong encoding

164+ , do k <- choose (0,0xFFFF)

165+ let c = chr k

166+ case k of

167+ _ | k < 0x80 -> oneof [ let (w,x) = ord2 c in return [w,x]

168+ , let (w,x,y) = ord3 c in return [w,x,y]

169+ , let (w,x,y,z) = ord4 c in return [w,x,y,z] ]

170+ | k < 0x7FF -> oneof [ let (w,x,y) = ord3 c in return [w,x,y]

171+ , let (w,x,y,z) = ord4 c in return [w,x,y,z] ]

172+ | otherwise -> let (w,x,y,z) = ord4 c in return [w,x,y,z]

173+ ]

174+ where

175+ contByte = (0x80 +) <$> choose (0, 0x3f)

176+ upTo n gen = do

177+ k <- choose (0,n)

178+ vectorOf k gen

179+

130180s_Eq s = (s==) `eq` ((S.streamList s==) . S.streamList)

131181 where _types = s :: String

132182sf_Eq p s =

@@ -147,8 +197,10 @@ t_Show = show `eq` (show . T.pack)

147197tl_Show = show `eq` (show . TL.pack)

148198t_mappend s = mappend s`eqP` (unpackS . mappend (T.pack s))

149199tl_mappend s = mappend s`eqP` (unpackS . mappend (TL.pack s))

150−t_mconcat = mconcat `eq` (unpackS . mconcat . L.map T.pack)

151−tl_mconcat = mconcat `eq` (unpackS . mconcat . L.map TL.pack)

200+t_mconcat = unsquare $

201+ mconcat `eq` (unpackS . mconcat . L.map T.pack)

202+tl_mconcat = unsquare $

203+ mconcat `eq` (unpackS . mconcat . L.map TL.pack)

152204t_mempty = mempty == (unpackS (mempty :: T.Text))

153205tl_mempty = mempty == (unpackS (mempty :: TL.Text))

154206t_IsString = fromString `eqP` (T.unpack . fromString)

@@ -212,11 +264,14 @@ s_map_s f = map f `eqP` (unpackS . S.unstream . S.map f)

212264sf_map p f = (map f . L.filter p) `eqP` (unpackS . S.map f . S.filter p)

213265t_map f = map f `eqP` (unpackS . T.map f)

214266tl_map f = map f `eqP` (unpackS . TL.map f)

215−s_intercalate c = L.intercalate c `eq`

267+s_intercalate c = unsquare $

268+ L.intercalate c `eq`

216269 (unpackS . S.intercalate (packS c) . map packS)

217−t_intercalate c = L.intercalate c `eq`

270+t_intercalate c = unsquare $

271+ L.intercalate c `eq`

218272 (unpackS . T.intercalate (packS c) . map packS)

219−tl_intercalate c = L.intercalate c `eq`

273+tl_intercalate c = unsquare $

274+ L.intercalate c `eq`

220275 (unpackS . TL.intercalate (TL.pack c) . map TL.pack)

221276s_intersperse c = L.intersperse c `eqP`

222277 (unpackS . S.intersperse c)

@@ -224,20 +279,20 @@ s_intersperse_s c = L.intersperse c `eqP`

224279 (unpackS . S.unstream . S.intersperse c)

225280sf_intersperse p c= (L.intersperse c . L.filter p) `eqP`

226281 (unpackS . S.intersperse c . S.filter p)

227−t_intersperse c = L.intersperse c `eqP` (unpackS . T.intersperse c)

228−tl_intersperse c = L.intersperse c `eqP` (unpackS . TL.intersperse c)

229−t_transpose = L.transpose `eq` (map unpackS . T.transpose . map packS)

230−tl_transpose = L.transpose `eq` (map unpackS . TL.transpose . map TL.pack)

282+t_intersperse c = unsquare $

283+ L.intersperse c `eqP` (unpackS . T.intersperse c)

284+tl_intersperse c = unsquare $

285+ L.intersperse c `eqP` (unpackS . TL.intersperse c)

286+t_transpose = unsquare $

287+ L.transpose `eq` (map unpackS . T.transpose . map packS)

288+tl_transpose = unsquare $

289+ L.transpose `eq` (map unpackS . TL.transpose . map TL.pack)

231290t_reverse = L.reverse `eqP` (unpackS . T.reverse)

232291tl_reverse = L.reverse `eqP` (unpackS . TL.reverse)

233292t_reverse_short n = L.reverse `eqP` (unpackS . S.reverse . shorten n . S.stream)

234293

235294t_replace s d = (L.intercalate d . splitOn s) `eqP`

236295 (unpackS . T.replace (T.pack s) (T.pack d))

237−t_replace2 (NotEmpty s0) (NotEmpty d0) (NotEmpty w0) =

238− T.replace s d w =^= T.replace2 s d w

239− where s = f s0; d = f d0; w = f w0

240− f = T.tail . T.pack

241296tl_replace s d = (L.intercalate d . splitOn s) `eqP`

242297 (unpackS . TL.replace (TL.pack s) (TL.pack d))

243298

@@ -328,17 +383,24 @@ sf_foldr p f z = (L.foldr f z . L.filter p) `eqP` (S.foldr f z . S.filter p)

328383 where _types = f :: Char -> Char -> Char

329384t_foldr f z = L.foldr f z `eqP` T.foldr f z

330385 where _types = f :: Char -> Char -> Char

331−tl_foldr f z = L.foldr f z `eqP` TL.foldr f z

386+tl_foldr f z = unsquare $

387+ L.foldr f z `eqP` TL.foldr f z

332388 where _types = f :: Char -> Char -> Char

333−sf_foldr1 p f = (L.foldr1 f . L.filter p) `eqP` (S.foldr1 f . S.filter p)

389+sf_foldr1 p f = unsquare $

390+ (L.foldr1 f . L.filter p) `eqP` (S.foldr1 f . S.filter p)

334391t_foldr1 f = L.foldr1 f `eqP` T.foldr1 f

335−tl_foldr1 f = L.foldr1 f `eqP` TL.foldr1 f

392+tl_foldr1 f = unsquare $

393+ L.foldr1 f `eqP` TL.foldr1 f

336394

337−s_concat_s = L.concat `eq` (unpackS . S.unstream . S.concat . map packS)

338−sf_concat p = (L.concat . map (L.filter p)) `eq`

395+s_concat_s = unsquare $

396+ L.concat `eq` (unpackS . S.unstream . S.concat . map packS)

397+sf_concat p = unsquare $

398+ (L.concat . map (L.filter p)) `eq`

339399 (unpackS . S.concat . map (S.filter p . packS))

340−t_concat = L.concat `eq` (unpackS . T.concat . map packS)

341−tl_concat = L.concat `eq` (unpackS . TL.concat . map TL.pack)

400+t_concat = unsquare $

401+ L.concat `eq` (unpackS . T.concat . map packS)

402+tl_concat = unsquare $

403+ L.concat `eq` (unpackS . TL.concat . map TL.pack)

342404sf_concatMap p f = unsquare $ (L.concatMap f . L.filter p) `eqP`

343405 (unpackS . S.concatMap (packS . f) . S.filter p)

344406t_concatMap f = unsquare $

@@ -378,6 +440,10 @@ t_mapAccumR f z = L.mapAccumR f z `eqP` (second unpackS . T.mapAccumR f z)

378440tl_mapAccumR f z = L.mapAccumR f z `eqP` (second unpackS . TL.mapAccumR f z)

379441 where _types = f :: Int -> Char -> (Int,Char)

380442

443+tl_repeat n = (L.take m . L.repeat) `eq`

444+ (unpackS . TL.take (fromIntegral m) . TL.repeat)

445+ where m = fromIntegral (n :: Word8)

446+

381447replicate n l = concat (L.replicate n l)

382448

383449s_replicate n = replicate m `eq`

@@ -389,6 +455,14 @@ tl_replicate n = replicate m `eq`

389455 (unpackS . TL.replicate (fromIntegral m) . packS)

390456 where m = fromIntegral (n :: Word8)

391457

458+tl_cycle n = (L.take m . L.cycle) `eq`

459+ (unpackS . TL.take (fromIntegral m) . TL.cycle . packS)

460+ where m = fromIntegral (n :: Word8)

461+

462+tl_iterate f n = (L.take m . L.iterate f) `eq`

463+ (unpackS . TL.take (fromIntegral m) . TL.iterate f)

464+ where m = fromIntegral (n :: Word8)

465+

392466unf :: Int -> Char -> Maybe (Char, Char)

393467unf n c | fromEnum c * 100 > n = Nothing

394468 | otherwise = Just (c, succ c)

@@ -415,14 +489,22 @@ s_take_s m = L.take n `eqP` (unpackS . S.unstream . S.take n)

415489sf_take p n = (L.take n . L.filter p) `eqP`

416490 (unpackS . S.take n . S.filter p)

417491t_take n = L.take n `eqP` (unpackS . T.take n)

492+t_takeEnd n = (L.reverse . L.take n . L.reverse) `eqP`

493+ (unpackS . T.takeEnd n)

418494tl_take n = L.take n `eqP` (unpackS . TL.take (fromIntegral n))

495+tl_takeEnd n = (L.reverse . L.take (fromIntegral n) . L.reverse) `eqP`

496+ (unpackS . TL.takeEnd n)

419497s_drop n = L.drop n `eqP` (unpackS . S.drop n)

420498s_drop_s m = L.drop n `eqP` (unpackS . S.unstream . S.drop n)

421499 where n = small m

422500sf_drop p n = (L.drop n . L.filter p) `eqP`

423501 (unpackS . S.drop n . S.filter p)

424502t_drop n = L.drop n `eqP` (unpackS . T.drop n)

503+t_dropEnd n = (L.reverse . L.drop n . L.reverse) `eqP`

504+ (unpackS . T.dropEnd n)

425505tl_drop n = L.drop n `eqP` (unpackS . TL.drop (fromIntegral n))

506+tl_dropEnd n = (L.reverse . L.drop n . L.reverse) `eqP`

507+ (unpackS . TL.dropEnd (fromIntegral n))

426508s_take_drop m = (L.take n . L.drop n) `eqP` (unpackS . S.take n . S.drop n)

427509 where n = small m

428510s_take_drop_s m = (L.take n . L.drop n) `eqP`

@@ -434,6 +516,10 @@ sf_takeWhile q p = (L.takeWhile p . L.filter q) `eqP`

434516 (unpackS . S.takeWhile p . S.filter q)

435517t_takeWhile p = L.takeWhile p `eqP` (unpackS . T.takeWhile p)

436518tl_takeWhile p = L.takeWhile p `eqP` (unpackS . TL.takeWhile p)

519+t_takeWhileEnd p = (L.reverse . L.takeWhile p . L.reverse) `eqP`

520+ (unpackS . T.takeWhileEnd p)

521+tl_takeWhileEnd p = (L.reverse . L.takeWhile p . L.reverse) `eqP`

522+ (unpackS . TL.takeWhileEnd p)

437523s_dropWhile p = L.dropWhile p `eqP` (unpackS . S.dropWhile p)

438524s_dropWhile_s p = L.dropWhile p `eqP` (unpackS . S.unstream . S.dropWhile p)

439525sf_dropWhile q p = (L.dropWhile p . L.filter q) `eqP`

@@ -486,24 +572,27 @@ tl_groupBy p = L.groupBy p `eqP` (map unpackS . TL.groupBy p)

486572t_inits = L.inits `eqP` (map unpackS . T.inits)

487573tl_inits = L.inits `eqP` (map unpackS . TL.inits)

488574t_tails = L.tails `eqP` (map unpackS . T.tails)

489−tl_tails = L.tails `eqP` (map unpackS . TL.tails)

490−t_findAppendId (NotEmpty s) = unsquare $ \ts ->

575+tl_tails = unsquare $

576+ L.tails `eqP` (map unpackS . TL.tails)

577+t_findAppendId = unsquare $ \(NotEmpty s) ts ->

491578 let t = T.intercalate s ts

492579 in all (==t) $ map (uncurry T.append) (T.breakOnAll s t)

493−tl_findAppendId (NotEmpty s) = unsquare $ \ts ->

580+tl_findAppendId = unsquare $ \(NotEmpty s) ts ->

494581 let t = TL.intercalate s ts

495582 in all (==t) $ map (uncurry TL.append) (TL.breakOnAll s t)

496−t_findContains (NotEmpty s) = all (T.isPrefixOf s . snd) . T.breakOnAll s .

497− T.intercalate s

498−tl_findContains (NotEmpty s) = all (TL.isPrefixOf s . snd) .

583+t_findContains = unsquare $ \(NotEmpty s) ->

584+ all (T.isPrefixOf s . snd) . T.breakOnAll s . T.intercalate s

585+tl_findContains = unsquare $ \(NotEmpty s) -> all (TL.isPrefixOf s . snd) .

499586 TL.breakOnAll s . TL.intercalate s

500587sl_filterCount c = (L.genericLength . L.filter (==c)) `eqP` SL.countChar c

501588t_findCount s = (L.length . T.breakOnAll s) `eq` T.count s

502589tl_findCount s = (L.genericLength . TL.breakOnAll s) `eq` TL.count s

503590

504−t_splitOn_split s = (T.splitOn s `eq` Slow.splitOn s) . T.intercalate s

505−tl_splitOn_split s = ((TL.splitOn (TL.fromStrict s) . TL.fromStrict) `eq`

506− (map TL.fromStrict . T.splitOn s)) . T.intercalate s

591+t_splitOn_split s = unsquare $

592+ (T.splitOn s `eq` Slow.splitOn s) . T.intercalate s

593+tl_splitOn_split s = unsquare $

594+ ((TL.splitOn (TL.fromStrict s) . TL.fromStrict) `eq`

595+ (map TL.fromStrict . T.splitOn s)) . T.intercalate s

507596t_splitOn_i (NotEmpty t) = id `eq` (T.intercalate t . T.splitOn t)

508597tl_splitOn_i (NotEmpty t) = id `eq` (TL.intercalate t . TL.splitOn t)

509598

@@ -545,10 +634,14 @@ t_lines' = lines' `eqP` (map unpackS . T.lines')

545634t_words = L.words `eqP` (map unpackS . T.words)

546635

547636tl_words = L.words `eqP` (map unpackS . TL.words)

548−t_unlines = L.unlines `eq` (unpackS . T.unlines . map packS)

549−tl_unlines = L.unlines `eq` (unpackS . TL.unlines . map packS)

550−t_unwords = L.unwords `eq` (unpackS . T.unwords . map packS)

551−tl_unwords = L.unwords `eq` (unpackS . TL.unwords . map packS)

637+t_unlines = unsquare $

638+ L.unlines `eq` (unpackS . T.unlines . map packS)

639+tl_unlines = unsquare $

640+ L.unlines `eq` (unpackS . TL.unlines . map packS)

641+t_unwords = unsquare $

642+ L.unwords `eq` (unpackS . T.unwords . map packS)

643+tl_unwords = unsquare $

644+ L.unwords `eq` (unpackS . TL.unwords . map packS)

552645

553646s_isPrefixOf s = L.isPrefixOf s `eqP`

554647 (S.isPrefixOf (S.stream $ packS s) . S.stream)

@@ -625,8 +718,9 @@ t_indices (NotEmpty s) = Slow.indices s `eq` indices s

625718tl_indices (NotEmpty s) = lazyIndices s `eq` S.indices s

626719 where lazyIndices ss t = map fromIntegral $ Slow.indices (conc ss) (conc t)

627720 conc = T.concat . TL.toChunks

628−t_indices_occurs (NotEmpty t) ts = let s = T.intercalate t ts

629− in Slow.indices t s == indices t s

721+t_indices_occurs = unsquare $ \(NotEmpty t) ts ->

722+ let s = T.intercalate t ts

723+ in Slow.indices t s == indices t s

630724

631725-- Bit shifts.

632726shiftL w = forAll (choose (0,width-1)) $ \k -> Bits.shiftL w k == U.shiftL w k

@@ -662,6 +756,7 @@ tb_decimal :: (Integral a, Show a) => a -> Bool

662756tb_decimal = (TB.toLazyText . TB.decimal) `eq` (TL.pack . show)

663757

664758tb_decimal_integer (a::Integer) = tb_decimal a

759+tb_decimal_integer_big (Big a) = tb_decimal a

665760tb_decimal_int (a::Int) = tb_decimal a

666761tb_decimal_int8 (a::Int8) = tb_decimal a

667762tb_decimal_int16 (a::Int16) = tb_decimal a

@@ -673,6 +768,11 @@ tb_decimal_word16 (a::Word16) = tb_decimal a

673768tb_decimal_word32 (a::Word32) = tb_decimal a

674769tb_decimal_word64 (a::Word64) = tb_decimal a

675770

771+tb_decimal_big_int (BigBounded (a::Int)) = tb_decimal a

772+tb_decimal_big_int64 (BigBounded (a::Int64)) = tb_decimal a

773+tb_decimal_big_word (BigBounded (a::Word)) = tb_decimal a

774+tb_decimal_big_word64 (BigBounded (a::Word64)) = tb_decimal a

775+

676776tb_hex :: (Integral a, Show a) => a -> Bool

677777tb_hex = (TB.toLazyText . TB.hexadecimal) `eq` (TL.pack . flip showHex "")

678778

@@ -694,6 +794,21 @@ tb_realfloat = (TB.toLazyText . TB.realFloat) `eq` (TL.pack . show)

694794tb_realfloat_float (a::Float) = tb_realfloat a

695795tb_realfloat_double (a::Double) = tb_realfloat a

696796

797+showFloat :: (RealFloat a) => TB.FPFormat -> Maybe Int -> a -> ShowS

798+showFloat TB.Exponent = showEFloat

799+showFloat TB.Fixed = showFFloat

800+showFloat TB.Generic = showGFloat

801+

802+tb_formatRealFloat :: (RealFloat a, Show a) =>

803+ a -> TB.FPFormat -> Precision a -> Property

804+tb_formatRealFloat a fmt prec =

805+ TB.formatRealFloat fmt p a ===

806+ TB.fromString (showFloat fmt p a "")

807+ where p = precision a prec

808+

809+tb_formatRealFloat_float (a::Float) = tb_formatRealFloat a

810+tb_formatRealFloat_double (a::Double) = tb_formatRealFloat a

811+

697812-- Reading.

698813

699814t_decimal (n::Int) s =

@@ -702,16 +817,18 @@ t_decimal (n::Int) s =

702817tl_decimal (n::Int) s =

703818 TL.signed TL.decimal (TL.pack (show n) `TL.append` t) == Right (n,t)

704819 where t = TL.dropWhile isDigit s

705−t_hexadecimal (n::Positive Int) s ox =

820+t_hexadecimal m s ox =

706821 T.hexadecimal (T.concat [p, T.pack (showHex n ""), t]) == Right (n,t)

707822 where t = T.dropWhile isHexDigit s

708823 p = if ox then "0x" else ""

709−tl_hexadecimal (n::Positive Int) s ox =

824+ n = getPositive m :: Int

825+tl_hexadecimal m s ox =

710826 TL.hexadecimal (TL.concat [p, TL.pack (showHex n ""), t]) == Right (n,t)

711827 where t = TL.dropWhile isHexDigit s

712828 p = if ox then "0x" else ""

829+ n = getPositive m :: Int

713830

714−isFloaty c = c `elem` "+-.0123456789eE"

831+isFloaty c = c `elem` ("+-.0123456789eE" :: String)

715832

716833t_read_rational p tol (n::Double) s =

717834 case p (T.pack (show n) `T.append` t) of

@@ -793,6 +910,7 @@ tests =

793910 testProperty "t_utf8" t_utf8,

794911 testProperty "t_utf8'" t_utf8',

795912 testProperty "t_utf8_incr" t_utf8_incr,

913+ testProperty "t_utf8_undecoded" t_utf8_undecoded,

796914 testProperty "tl_utf8" tl_utf8,

797915 testProperty "tl_utf8'" tl_utf8',

798916 testProperty "t_utf16LE" t_utf16LE,

@@ -900,7 +1018,6 @@ tests =

9001018 testProperty "tl_reverse" tl_reverse,

9011019 testProperty "t_reverse_short" t_reverse_short,

9021020 testProperty "t_replace" t_replace,

903− testProperty "t_replace2" t_replace2,

9041021 testProperty "tl_replace" tl_replace,

9051022

9061023 testGroup "case conversion" [

@@ -993,9 +1110,12 @@ tests =

9931110 ],

9941111

9951112 testGroup "unfolds" [

1113+ testProperty "tl_repeat" tl_repeat,

9961114 testProperty "s_replicate" s_replicate,

9971115 testProperty "t_replicate" t_replicate,

9981116 testProperty "tl_replicate" tl_replicate,

1117+ testProperty "tl_cycle" tl_cycle,

1118+ testProperty "tl_iterate" tl_iterate,

9991119 testProperty "t_unfoldr" t_unfoldr,

10001120 testProperty "tl_unfoldr" tl_unfoldr,

10011121 testProperty "t_unfoldrN" t_unfoldrN,

@@ -1009,12 +1129,16 @@ tests =

10091129 testProperty "s_take_s" s_take_s,

10101130 testProperty "sf_take" sf_take,

10111131 testProperty "t_take" t_take,

1132+ testProperty "t_takeEnd" t_takeEnd,

10121133 testProperty "tl_take" tl_take,

1134+ testProperty "tl_takeEnd" tl_takeEnd,

10131135 testProperty "s_drop" s_drop,

10141136 testProperty "s_drop_s" s_drop_s,

10151137 testProperty "sf_drop" sf_drop,

10161138 testProperty "t_drop" t_drop,

1139+ testProperty "t_dropEnd" t_dropEnd,

10171140 testProperty "tl_drop" tl_drop,

1141+ testProperty "tl_dropEnd" tl_dropEnd,

10181142 testProperty "s_take_drop" s_take_drop,

10191143 testProperty "s_take_drop_s" s_take_drop_s,

10201144 testProperty "s_takeWhile" s_takeWhile,

@@ -1022,6 +1146,8 @@ tests =

10221146 testProperty "sf_takeWhile" sf_takeWhile,

10231147 testProperty "t_takeWhile" t_takeWhile,

10241148 testProperty "tl_takeWhile" tl_takeWhile,

1149+ testProperty "t_takeWhileEnd" t_takeWhileEnd,

1150+ testProperty "tl_takeWhileEnd" tl_takeWhileEnd,

10251151 testProperty "sf_dropWhile" sf_dropWhile,

10261152 testProperty "s_dropWhile" s_dropWhile,

10271153 testProperty "s_dropWhile_s" s_dropWhile_s,

@@ -1167,11 +1293,16 @@ tests =

11671293 testProperty "tb_decimal_int32" tb_decimal_int32,

11681294 testProperty "tb_decimal_int64" tb_decimal_int64,

11691295 testProperty "tb_decimal_integer" tb_decimal_integer,

1296+ testProperty "tb_decimal_integer_big" tb_decimal_integer_big,

11701297 testProperty "tb_decimal_word" tb_decimal_word,

11711298 testProperty "tb_decimal_word8" tb_decimal_word8,

11721299 testProperty "tb_decimal_word16" tb_decimal_word16,

11731300 testProperty "tb_decimal_word32" tb_decimal_word32,

1174− testProperty "tb_decimal_word64" tb_decimal_word64

1301+ testProperty "tb_decimal_word64" tb_decimal_word64,

1302+ testProperty "tb_decimal_big_int" tb_decimal_big_int,

1303+ testProperty "tb_decimal_big_word" tb_decimal_big_word,

1304+ testProperty "tb_decimal_big_int64" tb_decimal_big_int64,

1305+ testProperty "tb_decimal_big_word64" tb_decimal_big_word64

11751306 ],

11761307 testGroup "hexadecimal" [

11771308 testProperty "tb_hexadecimal_int" tb_hexadecimal_int,

@@ -1188,7 +1319,9 @@ tests =

11881319 ],

11891320 testGroup "realfloat" [

11901321 testProperty "tb_realfloat_double" tb_realfloat_double,

1191− testProperty "tb_realfloat_float" tb_realfloat_float

1322+ testProperty "tb_realfloat_float" tb_realfloat_float,

1323+ testProperty "tb_formatRealFloat_float" tb_formatRealFloat_float,

1324+ testProperty "tb_formatRealFloat_double" tb_formatRealFloat_double

11921325 ],

11931326 testProperty "tb_fromText" tb_fromText,

11941327 testProperty "tb_singleton" tb_singleton

@@ -1224,5 +1357,7 @@ tests =

12241357 testProperty "t_take_drop_16" t_take_drop_16,

12251358 testProperty "t_use_from" t_use_from,

12261359 testProperty "t_copy" t_copy

1227− ]

1360+ ],

1361+

1362+ testGroup "mul" Mul.tests

12281363 ]

tests/Tests/Properties/Mul.hsnew file+40

@@ -0,0 +1,40 @@

1+{-# LANGUAGE ScopedTypeVariables #-}

3+module Tests.Properties.Mul (tests) where

5+import Control.Applicative ((<$>), pure)

6+import Control.Exception as E (SomeException, catch, evaluate)

7+import Data.Int (Int32, Int64)

8+import Data.Text.Internal (mul, mul32, mul64)

9+import System.IO.Unsafe (unsafePerformIO)

10+import Test.Framework (Test)

11+import Test.Framework.Providers.QuickCheck2 (testProperty)

12+import Test.QuickCheck hiding ((.&.))

13+

14+mulRef :: (Integral a, Bounded a) => a -> a -> Maybe a

15+mulRef a b

16+ | ab < bot || ab > top = Nothing

17+ | otherwise = Just (fromIntegral ab)

18+ where ab = fromIntegral a * fromIntegral b

19+ top = fromIntegral (maxBound `asTypeOf` a) :: Integer

20+ bot = fromIntegral (minBound `asTypeOf` a) :: Integer

21+

22+eval :: (a -> b -> c) -> a -> b -> Maybe c

23+eval f a b = unsafePerformIO $

24+ (Just <$> evaluate (f a b)) `E.catch` (\(_::SomeException) -> pure Nothing)

25+

26+t_mul32 :: Int32 -> Int32 -> Property

27+t_mul32 a b = mulRef a b === eval mul32 a b

28+

29+t_mul64 :: Int64 -> Int64 -> Property

30+t_mul64 a b = mulRef a b === eval mul64 a b

31+

32+t_mul :: Int -> Int -> Property

33+t_mul a b = mulRef a b === eval mul a b

34+

35+tests :: [Test]

36+tests = [

37+ testProperty "t_mul" t_mul

38+ , testProperty "t_mul32" t_mul32

39+ , testProperty "t_mul64" t_mul64

40+ ]

tests/Tests/QuickCheckUtils.hs+107−84

@@ -10,107 +10,91 @@ module Tests.QuickCheckUtils

1010 , unsquare

1111 , smallArbitrary

1212

13− , NotEmpty (..)

13+ , BigBounded(..)

14+ , BigInt(..)

15+ , NotEmpty(..)

1416

15− , Small (..)

17+ , Small(..)

1618 , small

1719

20+ , Precision(..)

21+ , precision

22+

1823 , integralRandomR

1924

20− , DecodeErr (..)

25+ , DecodeErr(..)

26+ , genDecodeErr

2127

22− , Stringy (..)

28+ , Stringy(..)

2329 , eq

2430 , eqP

2531

26− , Encoding (..)

32+ , Encoding(..)

2733

2834 , write_read

2935 ) where

3036

37+import Control.Applicative ((<$>))

3138import Control.Arrow (first, (***))

3239import Control.DeepSeq (NFData (..), deepseq)

3340import Control.Exception (bracket)

34−import Data.Bits ((.&.))

35−import Data.Char (chr)

36−import Data.Word (Word8, Word16)

3741import Data.String (IsString, fromString)

3842import Data.Text.Foreign (I16)

43+import Data.Text.Lazy.Builder.RealFloat (FPFormat(..))

44+import Data.Word (Word8, Word16)

3945import Debug.Trace (trace)

40−import System.Random (Random (..), RandomGen)

41−import Test.QuickCheck hiding ((.&.))

46+import System.Random (Random(..), RandomGen)

47+import Test.QuickCheck hiding (Fixed(..), Small (..), (.&.))

4248import Test.QuickCheck.Monadic (assert, monadicIO, run)

49+import Test.QuickCheck.Unicode (string)

50+import Tests.Utils

4351import qualified Data.ByteString as B

4452import qualified Data.Text as T

4553import qualified Data.Text.Encoding.Error as T

46−import qualified Data.Text.Fusion as TF

47−import qualified Data.Text.Fusion.Common as TF

54+import qualified Data.Text.Internal.Fusion as TF

55+import qualified Data.Text.Internal.Fusion.Common as TF

56+import qualified Data.Text.Internal.Lazy as TL

57+import qualified Data.Text.Internal.Lazy.Fusion as TLF

4858import qualified Data.Text.Lazy as TL

49−import qualified Data.Text.Lazy.Fusion as TLF

50−import qualified Data.Text.Lazy.Internal as TL

5159import qualified System.IO as IO

5260

53−import Tests.Utils

61+#if !MIN_VERSION_base(4,4,0)

62+import Data.Int (Int64)

63+import Data.Word (Word, Word64)

64+#endif

65+

66+genUnicode :: IsString a => Gen a

67+genUnicode = fromString <$> string

5468

5569instance Random I16 where

5670 randomR = integralRandomR

5771 random = randomR (minBound,maxBound)

5872

5973instance Arbitrary I16 where

60− arbitrary = choose (minBound,maxBound)

74+ arbitrary = arbitrarySizedIntegral

75+ shrink = shrinkIntegral

6176

6277instance Arbitrary B.ByteString where

6378 arbitrary = B.pack `fmap` arbitrary

79+ shrink = map B.pack . shrink . B.unpack

6480

65−genUnicode :: IsString a => Gen a

66−genUnicode = fmap fromString string where

67− string = sized $ \n ->

68− do k <- choose (0,n)

69− sequence [ char | _ <- [1..k] ]

70−

71− excluding :: [a -> Bool] -> Gen a -> Gen a

72− excluding bad gen = loop

73− where

74− loop = do

75− x <- gen

76− if or (map ($ x) bad)

77− then loop

78− else return x

79−

80− reserved = [lowSurrogate, highSurrogate, noncharacter]

81− lowSurrogate c = c >= 0xDC00 && c <= 0xDFFF

82− highSurrogate c = c >= 0xD800 && c <= 0xDBFF

83− noncharacter c = masked == 0xFFFE || masked == 0xFFFF

84− where

85− masked = c .&. 0xFFFF

86−

87− ascii = choose (0,0x7F)

88− plane0 = choose (0xF0, 0xFFFF)

89− plane1 = oneof [ choose (0x10000, 0x10FFF)

90− , choose (0x11000, 0x11FFF)

91− , choose (0x12000, 0x12FFF)

92− , choose (0x13000, 0x13FFF)

93− , choose (0x1D000, 0x1DFFF)

94− , choose (0x1F000, 0x1FFFF)

95− ]

96− plane2 = oneof [ choose (0x20000, 0x20FFF)

97− , choose (0x21000, 0x21FFF)

98− , choose (0x22000, 0x22FFF)

99− , choose (0x23000, 0x23FFF)

100− , choose (0x24000, 0x24FFF)

101− , choose (0x25000, 0x25FFF)

102− , choose (0x26000, 0x26FFF)

103− , choose (0x27000, 0x27FFF)

104− , choose (0x28000, 0x28FFF)

105− , choose (0x29000, 0x29FFF)

106− , choose (0x2A000, 0x2AFFF)

107− , choose (0x2B000, 0x2BFFF)

108− , choose (0x2F000, 0x2FFFF)

109− ]

110− plane14 = choose (0xE0000, 0xE0FFF)

111− planes = [ascii, plane0, plane1, plane2, plane14]

112−

113− char = chr `fmap` excluding reserved (oneof planes)

81+#if !MIN_VERSION_base(4,4,0)

82+instance Random Int64 where

83+ randomR = integralRandomR

84+ random = randomR (minBound,maxBound)

85+

86+instance Random Word where

87+ randomR = integralRandomR

88+ random = randomR (minBound,maxBound)

89+

90+instance Random Word8 where

91+ randomR = integralRandomR

92+ random = randomR (minBound,maxBound)

93+

94+instance Random Word64 where

95+ randomR = integralRandomR

96+ random = randomR (minBound,maxBound)

97+#endif

11498

11599-- For tests that have O(n^2) running times or input sizes, resize

116100-- their inputs to the square root of the originals.

@@ -123,9 +107,25 @@ smallArbitrary = sized $ \n -> resize (smallish n) arbitrary

123107

124108instance Arbitrary T.Text where

125109 arbitrary = T.pack `fmap` arbitrary

110+ shrink = map T.pack . shrink . T.unpack

126111

127112instance Arbitrary TL.Text where

128113 arbitrary = (TL.fromChunks . map notEmpty) `fmap` smallArbitrary

114+ shrink = map TL.pack . shrink . TL.unpack

115+

116+newtype BigInt = Big Integer

117+ deriving (Eq, Show)

118+

119+instance Arbitrary BigInt where

120+ arbitrary = choose (1::Int,200) >>= \e -> Big <$> choose (10^(e-1),10^e)

121+ shrink (Big a) = [Big (a `div` 2^(l-e)) | e <- shrink l]

122+ where l = truncate (log (fromIntegral a) / log 2 :: Double) :: Integer

123+

124+newtype BigBounded a = BigBounded a

125+ deriving (Eq, Show)

126+

127+instance (Bounded a, Random a, Arbitrary a) => Arbitrary (BigBounded a) where

128+ arbitrary = BigBounded <$> choose (minBound, maxBound)

129129

130130newtype NotEmpty a = NotEmpty { notEmpty :: a }

131131 deriving (Eq, Ord)

@@ -138,15 +138,23 @@ instance Functor NotEmpty where

138138

139139instance Arbitrary a => Arbitrary (NotEmpty [a]) where

140140 arbitrary = sized (\n -> NotEmpty `fmap` (choose (1,n+1) >>= vector))

141+ shrink = shrinkNotEmpty null

141142

142143instance Arbitrary (NotEmpty T.Text) where

143144 arbitrary = (fmap T.pack) `fmap` arbitrary

145+ shrink = shrinkNotEmpty T.null

144146

145147instance Arbitrary (NotEmpty TL.Text) where

146148 arbitrary = (fmap TL.pack) `fmap` arbitrary

149+ shrink = shrinkNotEmpty TL.null

147150

148151instance Arbitrary (NotEmpty B.ByteString) where

149152 arbitrary = (fmap B.pack) `fmap` arbitrary

153+ shrink = shrinkNotEmpty B.null

154+

155+shrinkNotEmpty :: Arbitrary a => (a -> Bool) -> NotEmpty a -> [NotEmpty a]

156+shrinkNotEmpty isNull (NotEmpty xs) =

157+ [ NotEmpty xs' | xs' <- shrink xs, not (isNull xs') ]

150158

151159data Small = S0 | S1 | S2 | S3 | S4 | S5 | S6 | S7

152160 | S8 | S9 | S10 | S11 | S12 | S13 | S14 | S15

@@ -187,23 +195,25 @@ instance Random Small where

187195 random = randomR (minBound,maxBound)

188196

189197instance Arbitrary Small where

190− arbitrary = choose (minBound,maxBound)

198+ arbitrary = choose (minBound, maxBound)

199+ shrink = shrinkIntegral

191200

192201integralRandomR :: (Integral a, RandomGen g) => (a,a) -> g -> (a,g)

193202integralRandomR (a,b) g = case randomR (fromIntegral a :: Integer,

194203 fromIntegral b :: Integer) g of

195204 (x,h) -> (fromIntegral x, h)

196205

197−data DecodeErr = DE String T.OnDecodeError

206+data DecodeErr = Lenient | Ignore | Strict | Replace

207+ deriving (Show, Eq)

198208

199−instance Show DecodeErr where

200− show (DE d _) = "DE " ++ d

209+genDecodeErr :: DecodeErr -> Gen T.OnDecodeError

210+genDecodeErr Lenient = return T.lenientDecode

211+genDecodeErr Ignore = return T.ignore

212+genDecodeErr Strict = return T.strictDecode

213+genDecodeErr Replace = arbitrary

201214

202215instance Arbitrary DecodeErr where

203− arbitrary = oneof [ return $ DE "lenient" T.lenientDecode

204− , return $ DE "ignore" T.ignore

205− , return $ DE "strict" T.strictDecode

206− , DE "replace" `fmap` arbitrary ]

216+ arbitrary = elements [Lenient, Ignore, Strict, Replace]

207217

208218class Stringy s where

209219 packS :: String -> s

@@ -257,6 +267,30 @@ eqP f g s w = eql "orig" (f s) (g t) &&

257267 | a =^= b = True

258268 | otherwise = trace (d ++ ": " ++ show a ++ " /= " ++ show b) False

259269

270+instance Arbitrary FPFormat where

271+ arbitrary = elements [Exponent, Fixed, Generic]

272+

273+newtype Precision a = Precision (Maybe Int)

274+ deriving (Eq, Show)

275+

276+precision :: a -> Precision a -> Maybe Int

277+precision _ (Precision prec) = prec

278+

279+arbitraryPrecision :: Int -> Gen (Precision a)

280+arbitraryPrecision maxDigits = Precision <$> do

281+ n <- choose (-1,maxDigits)

282+ return $ if n == -1

283+ then Nothing

284+ else Just n

285+

286+instance Arbitrary (Precision Float) where

287+ arbitrary = arbitraryPrecision 11

288+ shrink = map Precision . shrink . precision undefined

289+

290+instance Arbitrary (Precision Double) where

291+ arbitrary = arbitraryPrecision 22

292+ shrink = map Precision . shrink . precision undefined

293+

260294-- Work around lack of Show instance for TextEncoding.

261295data Encoding = E String IO.TextEncoding

262296

@@ -274,17 +308,6 @@ windowsNewlineMode = IO.NewlineMode

274308 { IO.inputNL = IO.CRLF, IO.outputNL = IO.CRLF

275309 }

276310

277−-- Newline and NewlineMode have standard Show instance from GHC 7 onwards

278−#if __GLASGOW_HASKELL__ < 700

279−instance Show IO.Newline where

280− show IO.CRLF = "CRLF"

281− show IO.LF = "LF"

282−

283−instance Show IO.NewlineMode where

284− show (IO.NewlineMode i o) = "NewlineMode { inputNL = " ++ show i ++

285− ", outputNL = " ++ show o ++ " }"

286−# endif

287−

288311instance Arbitrary IO.NewlineMode where

289312 arbitrary = oneof . map return $

290313 [ IO.noNewlineTranslation, IO.universalNewlineMode, IO.nativeNewlineMode

tests/cabal.confignew file+6

@@ -0,0 +1,6 @@

1+-- These flags help to speed up building the test suite.

3+documentation: False

4+executable-stripping: False

5+flags: developer

6+library-profiling: False

tests/literal-rule-test.shnew file+29

@@ -0,0 +1,29 @@

1+#!/bin/bash -e

3+failed=0

5+function check_firings() {

6+ rule=$1

7+ expected=$2

8+ build="ghc -O -ddump-rule-firings LiteralRuleTest.hs"

9+ build="$build -i.. -I../include"

10+ touch LiteralRuleTest.hs

11+ echo -n "Want to see $expected firings of rule $rule... " >&2

12+ firings=$($build 2>&1 | grep "Rule fired: $rule\$" | wc -l)

13+ rm -f LiteralRuleTest.{o.hi}

14+

15+ if [ $firings != $expected ]; then

16+ echo "failed, saw $firings" >&2

17+ failed=1

18+ else

19+ echo "pass" >&2

20+ fi

21+}

22+

23+check_firings "TEXT literal" 8

24+check_firings "TEXT literal UTF8" 7

25+check_firings "TEXT empty literal" 4

26+# This is broken at the moment. "TEXT literal" rule fires instead.

27+#check_firings "TEXT singleton literal" 5

28+

29+exit $failed

tests/text-tests.cabal+45−28

@@ -2,8 +2,8 @@ name: text-tests

22version: 0.0.0.0

33synopsis: Functional tests for the text package

44description: Functional tests for the text package

5−homepage: https://bitbucket.org/bos/text

6−license: BSD3

5+homepage: https://github.com/bos/text

6+license: BSD2

77license-file: ../LICENSE

88author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

99 Bryan O'Sullivan <bos@serpentine.com>,

@@ -18,10 +18,20 @@ cabal-version: >=1.8

1818flag hpc

1919 description: Enable HPC to generate coverage reports

2020 default: False

21+ manual: True

2122

2223executable text-tests

2324 main-is: Tests.hs

2425

26+ other-modules:

27+ Tests.IO

28+ Tests.Properties

29+ Tests.Properties.Mul

30+ Tests.QuickCheckUtils

31+ Tests.Regressions

32+ Tests.SlowFunctions

33+ Tests.Utils

34+

2535 ghc-options:

2636 -Wall -threaded -O0 -rtsopts

2737

@@ -30,16 +40,18 @@ executable text-tests

3040 -fhpc

3141

3242 cpp-options:

43+ -DTEST_SUITE

3344 -DASSERTS

3445 -DHAVE_DEEPSEQ

3546

3647 build-depends:

3748 HUnit >= 1.2,

38− QuickCheck >= 2.4,

49+ QuickCheck >= 2.7,

3950 base == 4.*,

4051 bytestring,

4152 deepseq,

4253 directory,

54+ quickcheck-unicode,

4355 random,

4456 test-framework >= 0.4,

4557 test-framework-hunit >= 0.2,

@@ -65,54 +77,58 @@ library

6577 hs-source-dirs: ..

6678 c-sources: ../cbits/cbits.c

6779 include-dirs: ../include

80+ ghc-options: -Wall

6881 exposed-modules:

6982 Data.Text

7083 Data.Text.Array

7184 Data.Text.Encoding

7285 Data.Text.Encoding.Error

73− Data.Text.Encoding.Fusion

74− Data.Text.Encoding.Fusion.Common

75− Data.Text.Encoding.Utf16

76− Data.Text.Encoding.Utf32

77− Data.Text.Encoding.Utf8

86+ Data.Text.Internal.Encoding.Fusion

87+ Data.Text.Internal.Encoding.Fusion.Common

88+ Data.Text.Internal.Encoding.Utf16

89+ Data.Text.Internal.Encoding.Utf32

90+ Data.Text.Internal.Encoding.Utf8

7891 Data.Text.Foreign

79− Data.Text.Fusion

80− Data.Text.Fusion.CaseMapping

81− Data.Text.Fusion.Common

82− Data.Text.Fusion.Internal

83− Data.Text.Fusion.Size

92+ Data.Text.Internal.Fusion

93+ Data.Text.Internal.Fusion.CaseMapping

94+ Data.Text.Internal.Fusion.Common

95+ Data.Text.Internal.Fusion.Size

96+ Data.Text.Internal.Fusion.Types

8497 Data.Text.IO

85− Data.Text.IO.Internal

98+ Data.Text.Internal.IO

8699 Data.Text.Internal

87100 Data.Text.Lazy

88101 Data.Text.Lazy.Builder

89− Data.Text.Lazy.Builder.Functions

102+ Data.Text.Internal.Builder.Functions

90103 Data.Text.Lazy.Builder.Int

91− Data.Text.Lazy.Builder.Int.Digits

92− Data.Text.Lazy.Builder.Internal

104+ Data.Text.Internal.Builder.Int.Digits

105+ Data.Text.Internal.Builder

93106 Data.Text.Lazy.Builder.RealFloat

94− Data.Text.Lazy.Builder.RealFloat.Functions

107+ Data.Text.Internal.Builder.RealFloat.Functions

95108 Data.Text.Lazy.Encoding

96− Data.Text.Lazy.Encoding.Fusion

97− Data.Text.Lazy.Fusion

109+ Data.Text.Internal.Lazy.Encoding.Fusion

110+ Data.Text.Internal.Lazy.Fusion

98111 Data.Text.Lazy.IO

99− Data.Text.Lazy.Internal

112+ Data.Text.Internal.Lazy

100113 Data.Text.Lazy.Read

101− Data.Text.Lazy.Search

102− Data.Text.Private

114+ Data.Text.Internal.Lazy.Search

115+ Data.Text.Internal.Private

103116 Data.Text.Read

104− Data.Text.Search

117+ Data.Text.Show

118+ Data.Text.Internal.Read

119+ Data.Text.Internal.Search

105120 Data.Text.Unsafe

106− Data.Text.Unsafe.Base

107− Data.Text.UnsafeChar

108− Data.Text.UnsafeShift

109− Data.Text.Util

121+ Data.Text.Internal.Unsafe

122+ Data.Text.Internal.Unsafe.Char

123+ Data.Text.Internal.Unsafe.Shift

124+ Data.Text.Internal.Functions

110125

111126 if flag(hpc)

112127 ghc-options:

113128 -fhpc

114129

115130 cpp-options:

131+ -DTEST_SUITE

116132 -DHAVE_DEEPSEQ

117133 -DASSERTS

118134 -DINTEGER_GMP

@@ -120,6 +136,7 @@ library

120136 build-depends:

121137 array,

122138 base == 4.*,

139+ binary,

123140 bytestring,

124141 deepseq,

125142 ghc-prim,

text.cabal+60−40

@@ -1,5 +1,5 @@

11name: text

2−version: 0.11.4.0

2+version: 1.2.2.1

33homepage: https://github.com/bos/text

44bug-reports: https://github.com/bos/text/issues

55synopsis: An efficient packed Unicode text type.

@@ -30,17 +30,8 @@ description:

3030 non-standard encodings, text breaking, and locales), see

3131 the @text-icu@ package:

3232 <http://hackage.haskell.org/package/text-icu>

33− .

34− —— RELEASE NOTES ——

35− .

36− Changes in 0.11.2.0:

37− .

38− * String literals are now converted directly from the format in

39− which GHC stores them into 'Text', without an intermediate

40− transformation through 'String', and without inlining of

41− conversion code at each site where a string literal is declared.

42− .

43−license: BSD3

33+

34+license: BSD2

4435license-file: LICENSE

4536author: Bryan O'Sullivan <bos@serpentine.com>

4637maintainer: Bryan O'Sullivan <bos@serpentine.com>

@@ -56,25 +47,37 @@ extra-source-files:

5647 benchmarks/cbits/*.c

5748 benchmarks/haskell/*.hs

5849 benchmarks/haskell/Benchmarks/*.hs

50+ benchmarks/haskell/Benchmarks/Programs/*.hs

5951 benchmarks/python/*.py

6052 benchmarks/ruby/*.rb

6153 benchmarks/text-benchmarks.cabal

54+ changelog.md

55+ include/*.h

6256 scripts/*.hs

6357 tests-and-benchmarks.markdown

6458 tests/*.hs

6559 tests/.ghci

6660 tests/Makefile

6761 tests/Tests/*.hs

62+ tests/Tests/Properties/*.hs

63+ tests/cabal.config

6864 tests/scripts/*.sh

6965 tests/text-tests.cabal

7066

67+flag bytestring-builder

68+ description: Depend on the bytestring-builder package for backwards compatibility.

69+ default: False

70+ manual: False

71+

7172flag developer

7273 description: operate in developer mode

7374 default: False

75+ manual: True

7476

7577flag integer-simple

7678 description: Use the simple integer library instead of GMP

7779 default: False

80+ manual: False

7881

7982library

8083 c-sources: cbits/cbits.c

@@ -88,6 +91,32 @@ library

8891 Data.Text.Foreign

8992 Data.Text.IO

9093 Data.Text.Internal

94+ Data.Text.Internal.Builder

95+ Data.Text.Internal.Builder.Functions

96+ Data.Text.Internal.Builder.Int.Digits

97+ Data.Text.Internal.Builder.RealFloat.Functions

98+ Data.Text.Internal.Encoding.Fusion

99+ Data.Text.Internal.Encoding.Fusion.Common

100+ Data.Text.Internal.Encoding.Utf16

101+ Data.Text.Internal.Encoding.Utf32

102+ Data.Text.Internal.Encoding.Utf8

103+ Data.Text.Internal.Functions

104+ Data.Text.Internal.Fusion

105+ Data.Text.Internal.Fusion.CaseMapping

106+ Data.Text.Internal.Fusion.Common

107+ Data.Text.Internal.Fusion.Size

108+ Data.Text.Internal.Fusion.Types

109+ Data.Text.Internal.IO

110+ Data.Text.Internal.Lazy

111+ Data.Text.Internal.Lazy.Encoding.Fusion

112+ Data.Text.Internal.Lazy.Fusion

113+ Data.Text.Internal.Lazy.Search

114+ Data.Text.Internal.Private

115+ Data.Text.Internal.Read

116+ Data.Text.Internal.Search

117+ Data.Text.Internal.Unsafe

118+ Data.Text.Internal.Unsafe.Char

119+ Data.Text.Internal.Unsafe.Shift

91120 Data.Text.Lazy

92121 Data.Text.Lazy.Builder

93122 Data.Text.Lazy.Builder.Int

@@ -98,39 +127,23 @@ library

98127 Data.Text.Lazy.Read

99128 Data.Text.Read

100129 Data.Text.Unsafe

130+

101131 other-modules:

102− Data.Text.Encoding.Fusion

103− Data.Text.Encoding.Fusion.Common

104− Data.Text.Encoding.Utf16

105− Data.Text.Encoding.Utf32

106− Data.Text.Encoding.Utf8

107− Data.Text.Fusion

108− Data.Text.Fusion.CaseMapping

109− Data.Text.Fusion.Common

110− Data.Text.Fusion.Internal

111− Data.Text.Fusion.Size

112− Data.Text.IO.Internal

113− Data.Text.Lazy.Builder.Functions

114− Data.Text.Lazy.Builder.Int.Digits

115− Data.Text.Lazy.Builder.Internal

116− Data.Text.Lazy.Builder.RealFloat.Functions

117− Data.Text.Lazy.Encoding.Fusion

118− Data.Text.Lazy.Fusion

119− Data.Text.Lazy.Search

120− Data.Text.Private

121− Data.Text.Search

122− Data.Text.Unsafe.Base

123− Data.Text.UnsafeChar

124− Data.Text.UnsafeShift

125− Data.Text.Util

132+ Data.Text.Show

126133

127134 build-depends:

128135 array >= 0.3,

129136 base >= 4.2 && < 5,

130− bytestring >= 0.9,

137+ binary,

131138 deepseq >= 1.1.0.0,

132139 ghc-prim >= 0.2

133140

141+ if flag(bytestring-builder)

142+ build-depends: bytestring >= 0.9 && < 0.10.4,

143+ bytestring-builder >= 0.10.4

144+ else

145+ build-depends: bytestring >= 0.10.4

146+

134147 cpp-options: -DHAVE_DEEPSEQ

135148 ghc-options: -Wall -fwarn-tabs -funbox-strict-fields -O2

136149 if flag(developer)

@@ -156,22 +169,29 @@ test-suite tests

156169 -Wall -threaded -O0 -rtsopts

157170

158171 cpp-options:

159− -DASSERTS -DHAVE_DEEPSEQ

172+ -DASSERTS -DHAVE_DEEPSEQ -DTEST_SUITE

160173

161174 build-depends:

162175 HUnit >= 1.2,

163− QuickCheck >= 2.4,

176+ QuickCheck >= 2.7,

164177 array,

165178 base,

166− bytestring,

179+ binary,

167180 deepseq,

168181 directory,

169182 ghc-prim,

183+ quickcheck-unicode,

170184 random,

171185 test-framework >= 0.4,

172186 test-framework-hunit >= 0.2,

173187 test-framework-quickcheck2 >= 0.2

174188

189+ if flag(bytestring-builder)

190+ build-depends: bytestring >= 0.9 && < 0.10.4,

191+ bytestring-builder >= 0.10.4

192+ else

193+ build-depends: bytestring >= 0.10.4

194+

175195 if flag(integer-simple)

176196 cpp-options: -DINTEGER_SIMPLE

177197 build-depends: integer-simple >= 0.1 && < 0.5